talk@lists.collectionspace.org

WE HAVE SUNSET THIS LISTSERV - Join us at collectionspace@lyrasislists.org

View all threads

Backup Practices of CollectionSpace community

MT
Ming Tsai
Mon, Dec 4, 2017 11:08 AM

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace for
our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated snapshots
of our linode instance (daily and weekly), with an occasional manual
snapshot. We do not have a local backup (actually, I haven't even looked
into how that might be done).

Would be great to hear some best practices that you have adopted. And, if
you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012

Dear Talklist Subscribers, My name is Ming Tsai. We are a recent implementer of Collectionspace for our family art and science foundation (tsaifoundation.org). With 60-70% of our inventory in Cspace now, I'd like to know how the general Cspace community handles the BACKUP of their database and their instance. Currently, our backup consists simply of having regular automated snapshots of our linode instance (daily and weekly), with an occasional manual snapshot. We do not have a local backup (actually, I haven't even looked into how that might be done). Would be great to hear some best practices that you have adopted. And, if you have specific methods, I'd be glad to see some procedures/steps. Thanks & best wishes, Ming Tsai Treasurer & CTO Tsai Art & Science Foundation New York, NY 10012
YN
Yousuf Nejati
Mon, Dec 4, 2017 1:06 PM

Hi Ming,

For backups my chosen methods are to (most of which you already know):

  1. Storing the latest source code for my CollectionSpace instance up in
    remote git repository, i.e. Github, Bitbucket, Gitlab, etc.
  2. Nightly PostgreSQL server database backups (you can create your own
    script or find some online)
  3. Nightly snapshots of the host server (again, you can create your own
    script or find some online)

Of course, if you're taking nightly snapshots of your host server, you
should be capturing the servers entire state, which should allow you to
recreate at any time.  One thing I would suggest would be to shut down
CollectionSpace and possibly your Postgres database (assuming it's only
being used for CollectionSpace) before taking the snapshot. You want to
make sure that nothing is being written to the CollectionSpace database and
the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're
taking your snapshot, otherwise you may find a few or many things that are
missing when it comes to recreate.  Finally, I would test that everything
is being backed up correctly by recreating your CollectionSpace instance in
a duplicate environment.

Best,

Yousuf

On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai mytsai2007@gmail.com wrote:

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace for
our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated
snapshots of our linode instance (daily and weekly), with an occasional
manual snapshot. We do not have a local backup (actually, I haven't even
looked into how that might be done).

Would be great to hear some best practices that you have adopted. And, if
you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012


Talk mailing list
Talk@lists.collectionspace.org
http://lists.collectionspace.org/mailman/listinfo/talk_
lists.collectionspace.org

--

Yousuf D. Nejati

Hi Ming, For backups my chosen methods are to (most of which you already know): 1) Storing the latest source code for my CollectionSpace instance up in remote git repository, i.e. Github, Bitbucket, Gitlab, etc. 2) Nightly PostgreSQL server database backups (you can create your own script or find some online) 3) Nightly snapshots of the host server (again, you can create your own script or find some online) Of course, if you're taking nightly snapshots of your host server, you should be capturing the servers entire state, which should allow you to recreate at any time. One thing I would suggest would be to shut down CollectionSpace and possibly your Postgres database (assuming it's only being used for CollectionSpace) before taking the snapshot. You want to make sure that nothing is being written to the CollectionSpace database and the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're taking your snapshot, otherwise you may find a few or many things that are missing when it comes to recreate. Finally, I would test that everything is being backed up correctly by recreating your CollectionSpace instance in a duplicate environment. Best, Yousuf On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai <mytsai2007@gmail.com> wrote: > Dear Talklist Subscribers, > > My name is Ming Tsai. We are a recent implementer of Collectionspace for > our family art and science foundation (tsaifoundation.org). > > With 60-70% of our inventory in Cspace now, I'd like to know how the > general Cspace community handles the BACKUP of their database and their > instance. > > Currently, our backup consists simply of having regular automated > snapshots of our linode instance (daily and weekly), with an occasional > manual snapshot. We do not have a local backup (actually, I haven't even > looked into how that might be done). > > Would be great to hear some best practices that you have adopted. And, if > you have specific methods, I'd be glad to see some procedures/steps. > > Thanks & best wishes, > Ming Tsai > > Treasurer & CTO > Tsai Art & Science Foundation > New York, NY 10012 > > _______________________________________________ > Talk mailing list > Talk@lists.collectionspace.org > http://lists.collectionspace.org/mailman/listinfo/talk_ > lists.collectionspace.org > > -- *Yousuf D. Nejati*
MT
Ming Tsai
Tue, Dec 5, 2017 2:55 PM

Hi Yousuf,

Nice to hear from you! and thanks for getting back to me on the backup
question.

I'm looking over some of my notes--it's been a long time since I've looked
at the backend of my instance--and I can't say I know which files make up
my 'source code.' Is it just the files in my .../tenants/ folders in the
'application' and 'ui' layers?

Would be great if you could tell me which folders contain my source code?

Had a quick look at the PostgreSQL website and there are apparently 3
methods of backup: SQL dump, file system level backup, and continuous
archiving. Which do you do? If you have a sample script that would be very
helpful.

As for the daily snapshot, it is an automatic Linode backup service, so I
wasn't thinking of shutting down Cspace and the postgres database daily. On
my occasional backup, however, I could certainly shutdown both programs
before taking the snapshot.

Thanks for any additional pointers.

With best wishes,
Ming Tsai

On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati yousufnejati@gmail.com
wrote:

Hi Ming,

For backups my chosen methods are to (most of which you already know):

  1. Storing the latest source code for my CollectionSpace instance up in
    remote git repository, i.e. Github, Bitbucket, Gitlab, etc.
  2. Nightly PostgreSQL server database backups (you can create your own
    script or find some online)
  3. Nightly snapshots of the host server (again, you can create your own
    script or find some online)

Of course, if you're taking nightly snapshots of your host server, you
should be capturing the servers entire state, which should allow you to
recreate at any time.  One thing I would suggest would be to shut down
CollectionSpace and possibly your Postgres database (assuming it's only
being used for CollectionSpace) before taking the snapshot. You want to
make sure that nothing is being written to the CollectionSpace database and
the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're
taking your snapshot, otherwise you may find a few or many things that are
missing when it comes to recreate.  Finally, I would test that everything
is being backed up correctly by recreating your CollectionSpace instance in
a duplicate environment.

Best,

Yousuf

On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai mytsai2007@gmail.com wrote:

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace for
our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated
snapshots of our linode instance (daily and weekly), with an occasional
manual snapshot. We do not have a local backup (actually, I haven't even
looked into how that might be done).

Would be great to hear some best practices that you have adopted. And, if
you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012


Talk mailing list
Talk@lists.collectionspace.org
http://lists.collectionspace.org/mailman/listinfo/talk_lists
.collectionspace.org

--

Yousuf D. Nejati

Hi Yousuf, Nice to hear from you! and thanks for getting back to me on the backup question. I'm looking over some of my notes--it's been a long time since I've looked at the backend of my instance--and I can't say I know which files make up my 'source code.' Is it just the files in my .../tenants/ folders in the 'application' and 'ui' layers? Would be great if you could tell me which folders contain my source code? Had a quick look at the PostgreSQL website and there are apparently 3 methods of backup: SQL dump, file system level backup, and continuous archiving. Which do you do? If you have a sample script that would be very helpful. As for the daily snapshot, it is an automatic Linode backup service, so I wasn't thinking of shutting down Cspace and the postgres database daily. On my occasional backup, however, I could certainly shutdown both programs before taking the snapshot. Thanks for any additional pointers. With best wishes, Ming Tsai On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati <yousufnejati@gmail.com> wrote: > Hi Ming, > > For backups my chosen methods are to (most of which you already know): > > 1) Storing the latest source code for my CollectionSpace instance up in > remote git repository, i.e. Github, Bitbucket, Gitlab, etc. > 2) Nightly PostgreSQL server database backups (you can create your own > script or find some online) > 3) Nightly snapshots of the host server (again, you can create your own > script or find some online) > > Of course, if you're taking nightly snapshots of your host server, you > should be capturing the servers entire state, which should allow you to > recreate at any time. One thing I would suggest would be to shut down > CollectionSpace and possibly your Postgres database (assuming it's only > being used for CollectionSpace) before taking the snapshot. You want to > make sure that nothing is being written to the CollectionSpace database and > the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're > taking your snapshot, otherwise you may find a few or many things that are > missing when it comes to recreate. Finally, I would test that everything > is being backed up correctly by recreating your CollectionSpace instance in > a duplicate environment. > > Best, > > Yousuf > > On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai <mytsai2007@gmail.com> wrote: > >> Dear Talklist Subscribers, >> >> My name is Ming Tsai. We are a recent implementer of Collectionspace for >> our family art and science foundation (tsaifoundation.org). >> >> With 60-70% of our inventory in Cspace now, I'd like to know how the >> general Cspace community handles the BACKUP of their database and their >> instance. >> >> Currently, our backup consists simply of having regular automated >> snapshots of our linode instance (daily and weekly), with an occasional >> manual snapshot. We do not have a local backup (actually, I haven't even >> looked into how that might be done). >> >> Would be great to hear some best practices that you have adopted. And, if >> you have specific methods, I'd be glad to see some procedures/steps. >> >> Thanks & best wishes, >> Ming Tsai >> >> Treasurer & CTO >> Tsai Art & Science Foundation >> New York, NY 10012 >> >> _______________________________________________ >> Talk mailing list >> Talk@lists.collectionspace.org >> http://lists.collectionspace.org/mailman/listinfo/talk_lists >> .collectionspace.org >> >> > > > -- > > *Yousuf D. Nejati* >
RL
Ray Lee
Tue, Dec 5, 2017 9:25 PM

Hi Ming,
At Berkeley our backup situation is somewhat esoteric, due to the number of
CollectionSpace instances we run, their size, and our use of services
offered by campus IT. For most installations I think Yousuf's
recommendations are spot on.

We do SQL dump (pg_dump) for the database. The tradeoff with filesystem
level backup is this: For filesystem level backup, the postgres server must
be stopped. For pg_dump, the database can be running. But filesystem level
backups can be restored just by restoring the filesystem, which is probably
more or less a button click in linode. SQL dumps require some additional
work on the command line to restore from the dump file.

For me the relevant questions are: Does linode allow you to schedule when
backups occur? Are you ok with having CSpace unusable around the time of
the backup?

Ray

On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai mytsai2007@gmail.com wrote:

Hi Yousuf,

Nice to hear from you! and thanks for getting back to me on the backup
question.

I'm looking over some of my notes--it's been a long time since I've looked
at the backend of my instance--and I can't say I know which files make up
my 'source code.' Is it just the files in my .../tenants/ folders in the
'application' and 'ui' layers?

Would be great if you could tell me which folders contain my source code?

Had a quick look at the PostgreSQL website and there are apparently 3
methods of backup: SQL dump, file system level backup, and continuous
archiving. Which do you do? If you have a sample script that would be very
helpful.

As for the daily snapshot, it is an automatic Linode backup service, so I
wasn't thinking of shutting down Cspace and the postgres database daily. On
my occasional backup, however, I could certainly shutdown both programs
before taking the snapshot.

Thanks for any additional pointers.

With best wishes,
Ming Tsai

On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati yousufnejati@gmail.com
wrote:

Hi Ming,

For backups my chosen methods are to (most of which you already know):

  1. Storing the latest source code for my CollectionSpace instance up in
    remote git repository, i.e. Github, Bitbucket, Gitlab, etc.
  2. Nightly PostgreSQL server database backups (you can create your own
    script or find some online)
  3. Nightly snapshots of the host server (again, you can create your own
    script or find some online)

Of course, if you're taking nightly snapshots of your host server, you
should be capturing the servers entire state, which should allow you to
recreate at any time.  One thing I would suggest would be to shut down
CollectionSpace and possibly your Postgres database (assuming it's only
being used for CollectionSpace) before taking the snapshot. You want to
make sure that nothing is being written to the CollectionSpace database and
the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're
taking your snapshot, otherwise you may find a few or many things that are
missing when it comes to recreate.  Finally, I would test that everything
is being backed up correctly by recreating your CollectionSpace instance in
a duplicate environment.

Best,

Yousuf

On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai mytsai2007@gmail.com wrote:

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace for
our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated
snapshots of our linode instance (daily and weekly), with an occasional
manual snapshot. We do not have a local backup (actually, I haven't even
looked into how that might be done).

Would be great to hear some best practices that you have adopted. And,
if you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012


Talk mailing list
Talk@lists.collectionspace.org
http://lists.collectionspace.org/mailman/listinfo/talk_lists
.collectionspace.org

--

Yousuf D. Nejati

Hi Ming, At Berkeley our backup situation is somewhat esoteric, due to the number of CollectionSpace instances we run, their size, and our use of services offered by campus IT. For most installations I think Yousuf's recommendations are spot on. We do SQL dump (pg_dump) for the database. The tradeoff with filesystem level backup is this: For filesystem level backup, the postgres server must be stopped. For pg_dump, the database can be running. But filesystem level backups can be restored just by restoring the filesystem, which is probably more or less a button click in linode. SQL dumps require some additional work on the command line to restore from the dump file. For me the relevant questions are: Does linode allow you to schedule when backups occur? Are you ok with having CSpace unusable around the time of the backup? Ray On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai <mytsai2007@gmail.com> wrote: > Hi Yousuf, > > Nice to hear from you! and thanks for getting back to me on the backup > question. > > I'm looking over some of my notes--it's been a long time since I've looked > at the backend of my instance--and I can't say I know which files make up > my 'source code.' Is it just the files in my .../tenants/ folders in the > 'application' and 'ui' layers? > > Would be great if you could tell me which folders contain my source code? > > Had a quick look at the PostgreSQL website and there are apparently 3 > methods of backup: SQL dump, file system level backup, and continuous > archiving. Which do you do? If you have a sample script that would be very > helpful. > > As for the daily snapshot, it is an automatic Linode backup service, so I > wasn't thinking of shutting down Cspace and the postgres database daily. On > my occasional backup, however, I could certainly shutdown both programs > before taking the snapshot. > > Thanks for any additional pointers. > > With best wishes, > Ming Tsai > > On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati <yousufnejati@gmail.com> > wrote: > >> Hi Ming, >> >> For backups my chosen methods are to (most of which you already know): >> >> 1) Storing the latest source code for my CollectionSpace instance up in >> remote git repository, i.e. Github, Bitbucket, Gitlab, etc. >> 2) Nightly PostgreSQL server database backups (you can create your own >> script or find some online) >> 3) Nightly snapshots of the host server (again, you can create your own >> script or find some online) >> >> Of course, if you're taking nightly snapshots of your host server, you >> should be capturing the servers entire state, which should allow you to >> recreate at any time. One thing I would suggest would be to shut down >> CollectionSpace and possibly your Postgres database (assuming it's only >> being used for CollectionSpace) before taking the snapshot. You want to >> make sure that nothing is being written to the CollectionSpace database and >> the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're >> taking your snapshot, otherwise you may find a few or many things that are >> missing when it comes to recreate. Finally, I would test that everything >> is being backed up correctly by recreating your CollectionSpace instance in >> a duplicate environment. >> >> Best, >> >> Yousuf >> >> On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai <mytsai2007@gmail.com> wrote: >> >>> Dear Talklist Subscribers, >>> >>> My name is Ming Tsai. We are a recent implementer of Collectionspace for >>> our family art and science foundation (tsaifoundation.org). >>> >>> With 60-70% of our inventory in Cspace now, I'd like to know how the >>> general Cspace community handles the BACKUP of their database and their >>> instance. >>> >>> Currently, our backup consists simply of having regular automated >>> snapshots of our linode instance (daily and weekly), with an occasional >>> manual snapshot. We do not have a local backup (actually, I haven't even >>> looked into how that might be done). >>> >>> Would be great to hear some best practices that you have adopted. And, >>> if you have specific methods, I'd be glad to see some procedures/steps. >>> >>> Thanks & best wishes, >>> Ming Tsai >>> >>> Treasurer & CTO >>> Tsai Art & Science Foundation >>> New York, NY 10012 >>> >>> _______________________________________________ >>> Talk mailing list >>> Talk@lists.collectionspace.org >>> http://lists.collectionspace.org/mailman/listinfo/talk_lists >>> .collectionspace.org >>> >>> >> >> >> -- >> >> *Yousuf D. Nejati* >> > > > _______________________________________________ > Talk mailing list > Talk@lists.collectionspace.org > http://lists.collectionspace.org/mailman/listinfo/talk_ > lists.collectionspace.org > >
MT
Ming Tsai
Wed, Dec 6, 2017 9:27 AM

Hi Ray,

Thanks for your input. Yes, totally okay with having CSpace off-line during
time of backup. Just checked my Linode and it gives me a choice of 2-hour
windows throughout the day where it'll do the daily backup.

Would be happy to know more about both the 'SQL dump' versus the
'filesystem level' approach. I'm somewhat comfortable with working the
command line though certainly no pro.

Am I correct in assuming that most people backup to 3 locations (the cloud
service, github, and locally on their machine)? For me, it would be very
nice to have the instance backed-up on Linode, Github, and occasionally on
my local machine for peace of mind!

Could you share some specific details/demo scripts on your approach?

Thanks and best wishes,
Ming

On Tue, Dec 5, 2017 at 10:25 PM, Ray Lee rhlee@berkeley.edu wrote:

Hi Ming,
At Berkeley our backup situation is somewhat esoteric, due to the number
of CollectionSpace instances we run, their size, and our use of services
offered by campus IT. For most installations I think Yousuf's
recommendations are spot on.

We do SQL dump (pg_dump) for the database. The tradeoff with filesystem
level backup is this: For filesystem level backup, the postgres server must
be stopped. For pg_dump, the database can be running. But filesystem level
backups can be restored just by restoring the filesystem, which is probably
more or less a button click in linode. SQL dumps require some additional
work on the command line to restore from the dump file.

For me the relevant questions are: Does linode allow you to schedule when
backups occur? Are you ok with having CSpace unusable around the time of
the backup?

Ray

On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai mytsai2007@gmail.com wrote:

Hi Yousuf,

Nice to hear from you! and thanks for getting back to me on the backup
question.

I'm looking over some of my notes--it's been a long time since I've
looked at the backend of my instance--and I can't say I know which files
make up my 'source code.' Is it just the files in my .../tenants/ folders
in the 'application' and 'ui' layers?

Would be great if you could tell me which folders contain my source code?

Had a quick look at the PostgreSQL website and there are apparently 3
methods of backup: SQL dump, file system level backup, and continuous
archiving. Which do you do? If you have a sample script that would be very
helpful.

As for the daily snapshot, it is an automatic Linode backup service, so I
wasn't thinking of shutting down Cspace and the postgres database daily. On
my occasional backup, however, I could certainly shutdown both programs
before taking the snapshot.

Thanks for any additional pointers.

With best wishes,
Ming Tsai

On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati yousufnejati@gmail.com
wrote:

Hi Ming,

For backups my chosen methods are to (most of which you already know):

  1. Storing the latest source code for my CollectionSpace instance up in
    remote git repository, i.e. Github, Bitbucket, Gitlab, etc.
  2. Nightly PostgreSQL server database backups (you can create your own
    script or find some online)
  3. Nightly snapshots of the host server (again, you can create your own
    script or find some online)

Of course, if you're taking nightly snapshots of your host server, you
should be capturing the servers entire state, which should allow you to
recreate at any time.  One thing I would suggest would be to shut down
CollectionSpace and possibly your Postgres database (assuming it's only
being used for CollectionSpace) before taking the snapshot. You want to
make sure that nothing is being written to the CollectionSpace database and
the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're
taking your snapshot, otherwise you may find a few or many things that are
missing when it comes to recreate.  Finally, I would test that everything
is being backed up correctly by recreating your CollectionSpace instance in
a duplicate environment.

Best,

Yousuf

On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai mytsai2007@gmail.com wrote:

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace
for our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated
snapshots of our linode instance (daily and weekly), with an occasional
manual snapshot. We do not have a local backup (actually, I haven't even
looked into how that might be done).

Would be great to hear some best practices that you have adopted. And,
if you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012


Talk mailing list
Talk@lists.collectionspace.org
http://lists.collectionspace.org/mailman/listinfo/talk_lists
.collectionspace.org

--

Yousuf D. Nejati

Hi Ray, Thanks for your input. Yes, totally okay with having CSpace off-line during time of backup. Just checked my Linode and it gives me a choice of 2-hour windows throughout the day where it'll do the daily backup. Would be happy to know more about both the 'SQL dump' versus the 'filesystem level' approach. I'm somewhat comfortable with working the command line though certainly no pro. Am I correct in assuming that most people backup to 3 locations (the cloud service, github, and locally on their machine)? For me, it would be very nice to have the instance backed-up on Linode, Github, and occasionally on my local machine for peace of mind! Could you share some specific details/demo scripts on your approach? Thanks and best wishes, Ming On Tue, Dec 5, 2017 at 10:25 PM, Ray Lee <rhlee@berkeley.edu> wrote: > Hi Ming, > At Berkeley our backup situation is somewhat esoteric, due to the number > of CollectionSpace instances we run, their size, and our use of services > offered by campus IT. For most installations I think Yousuf's > recommendations are spot on. > > We do SQL dump (pg_dump) for the database. The tradeoff with filesystem > level backup is this: For filesystem level backup, the postgres server must > be stopped. For pg_dump, the database can be running. But filesystem level > backups can be restored just by restoring the filesystem, which is probably > more or less a button click in linode. SQL dumps require some additional > work on the command line to restore from the dump file. > > For me the relevant questions are: Does linode allow you to schedule when > backups occur? Are you ok with having CSpace unusable around the time of > the backup? > > Ray > > On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai <mytsai2007@gmail.com> wrote: > >> Hi Yousuf, >> >> Nice to hear from you! and thanks for getting back to me on the backup >> question. >> >> I'm looking over some of my notes--it's been a long time since I've >> looked at the backend of my instance--and I can't say I know which files >> make up my 'source code.' Is it just the files in my .../tenants/ folders >> in the 'application' and 'ui' layers? >> >> Would be great if you could tell me which folders contain my source code? >> >> Had a quick look at the PostgreSQL website and there are apparently 3 >> methods of backup: SQL dump, file system level backup, and continuous >> archiving. Which do you do? If you have a sample script that would be very >> helpful. >> >> As for the daily snapshot, it is an automatic Linode backup service, so I >> wasn't thinking of shutting down Cspace and the postgres database daily. On >> my occasional backup, however, I could certainly shutdown both programs >> before taking the snapshot. >> >> Thanks for any additional pointers. >> >> With best wishes, >> Ming Tsai >> >> On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati <yousufnejati@gmail.com> >> wrote: >> >>> Hi Ming, >>> >>> For backups my chosen methods are to (most of which you already know): >>> >>> 1) Storing the latest source code for my CollectionSpace instance up in >>> remote git repository, i.e. Github, Bitbucket, Gitlab, etc. >>> 2) Nightly PostgreSQL server database backups (you can create your own >>> script or find some online) >>> 3) Nightly snapshots of the host server (again, you can create your own >>> script or find some online) >>> >>> Of course, if you're taking nightly snapshots of your host server, you >>> should be capturing the servers entire state, which should allow you to >>> recreate at any time. One thing I would suggest would be to shut down >>> CollectionSpace and possibly your Postgres database (assuming it's only >>> being used for CollectionSpace) before taking the snapshot. You want to >>> make sure that nothing is being written to the CollectionSpace database and >>> the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're >>> taking your snapshot, otherwise you may find a few or many things that are >>> missing when it comes to recreate. Finally, I would test that everything >>> is being backed up correctly by recreating your CollectionSpace instance in >>> a duplicate environment. >>> >>> Best, >>> >>> Yousuf >>> >>> On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai <mytsai2007@gmail.com> wrote: >>> >>>> Dear Talklist Subscribers, >>>> >>>> My name is Ming Tsai. We are a recent implementer of Collectionspace >>>> for our family art and science foundation (tsaifoundation.org). >>>> >>>> With 60-70% of our inventory in Cspace now, I'd like to know how the >>>> general Cspace community handles the BACKUP of their database and their >>>> instance. >>>> >>>> Currently, our backup consists simply of having regular automated >>>> snapshots of our linode instance (daily and weekly), with an occasional >>>> manual snapshot. We do not have a local backup (actually, I haven't even >>>> looked into how that might be done). >>>> >>>> Would be great to hear some best practices that you have adopted. And, >>>> if you have specific methods, I'd be glad to see some procedures/steps. >>>> >>>> Thanks & best wishes, >>>> Ming Tsai >>>> >>>> Treasurer & CTO >>>> Tsai Art & Science Foundation >>>> New York, NY 10012 >>>> >>>> _______________________________________________ >>>> Talk mailing list >>>> Talk@lists.collectionspace.org >>>> http://lists.collectionspace.org/mailman/listinfo/talk_lists >>>> .collectionspace.org >>>> >>>> >>> >>> >>> -- >>> >>> *Yousuf D. Nejati* >>> >> >> >> _______________________________________________ >> Talk mailing list >> Talk@lists.collectionspace.org >> http://lists.collectionspace.org/mailman/listinfo/talk_lists >> .collectionspace.org >> >> >
RL
Ray Lee
Thu, Dec 7, 2017 11:49 PM

Hi Ming,
We don't directly deal with database backups any more, because we use a
campus service. I'm attaching the script we used when we did, although it's
not exactly what I'd use today, and it contains some things that are
specific to our installation. Also see this issue for how we used to do the
restore, but this may be outdated and/or depend on specifics of our
installation: https://issues.collectionspace.org/browse/PAHMA-388

These docs have more information about filesystem backup vs. SQL dump:
https://www.postgresql.org/docs/9.6/static/backup-file.html
https://www.postgresql.org/docs/9.6/static/backup-dump.html

Generally you want two backups:

  1. A local backup -- in this case, local to Linode
  2. A remote backup at a different geography/vendor

There are three things you need to back up in a CollectionSpace system:
a. Your configuration and customizations (aka source code)
b. Your postgres databases
c. Your uploaded data (nuxeo-server/data)

A Linode backup/snapshot takes care of 1a, 1b, and 1c.

You can use GitHub for 2a, but you don't have to. It's common because lots
of us already use it, and it's good to have some kind of source control in
any case. 2b and 2c will vary. You might use another cloud backup service.
You might write some scripts that transfer the files to your own computer.

Ray

On Wed, Dec 6, 2017 at 1:27 AM, Ming Tsai mytsai2007@gmail.com wrote:

Hi Ray,

Thanks for your input. Yes, totally okay with having CSpace off-line
during time of backup. Just checked my Linode and it gives me a choice of
2-hour windows throughout the day where it'll do the daily backup.

Would be happy to know more about both the 'SQL dump' versus the
'filesystem level' approach. I'm somewhat comfortable with working the
command line though certainly no pro.

Am I correct in assuming that most people backup to 3 locations (the cloud
service, github, and locally on their machine)? For me, it would be very
nice to have the instance backed-up on Linode, Github, and occasionally on
my local machine for peace of mind!

Could you share some specific details/demo scripts on your approach?

Thanks and best wishes,
Ming

On Tue, Dec 5, 2017 at 10:25 PM, Ray Lee rhlee@berkeley.edu wrote:

Hi Ming,
At Berkeley our backup situation is somewhat esoteric, due to the number
of CollectionSpace instances we run, their size, and our use of services
offered by campus IT. For most installations I think Yousuf's
recommendations are spot on.

We do SQL dump (pg_dump) for the database. The tradeoff with filesystem
level backup is this: For filesystem level backup, the postgres server must
be stopped. For pg_dump, the database can be running. But filesystem level
backups can be restored just by restoring the filesystem, which is probably
more or less a button click in linode. SQL dumps require some additional
work on the command line to restore from the dump file.

For me the relevant questions are: Does linode allow you to schedule when
backups occur? Are you ok with having CSpace unusable around the time of
the backup?

Ray

On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai mytsai2007@gmail.com wrote:

Hi Yousuf,

Nice to hear from you! and thanks for getting back to me on the backup
question.

I'm looking over some of my notes--it's been a long time since I've
looked at the backend of my instance--and I can't say I know which files
make up my 'source code.' Is it just the files in my .../tenants/ folders
in the 'application' and 'ui' layers?

Would be great if you could tell me which folders contain my source code?

Had a quick look at the PostgreSQL website and there are apparently 3
methods of backup: SQL dump, file system level backup, and continuous
archiving. Which do you do? If you have a sample script that would be very
helpful.

As for the daily snapshot, it is an automatic Linode backup service, so
I wasn't thinking of shutting down Cspace and the postgres database daily.
On my occasional backup, however, I could certainly shutdown both programs
before taking the snapshot.

Thanks for any additional pointers.

With best wishes,
Ming Tsai

On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati yousufnejati@gmail.com
wrote:

Hi Ming,

For backups my chosen methods are to (most of which you already know):

  1. Storing the latest source code for my CollectionSpace instance up in
    remote git repository, i.e. Github, Bitbucket, Gitlab, etc.
  2. Nightly PostgreSQL server database backups (you can create your own
    script or find some online)
  3. Nightly snapshots of the host server (again, you can create your own
    script or find some online)

Of course, if you're taking nightly snapshots of your host server, you
should be capturing the servers entire state, which should allow you to
recreate at any time.  One thing I would suggest would be to shut down
CollectionSpace and possibly your Postgres database (assuming it's only
being used for CollectionSpace) before taking the snapshot. You want to
make sure that nothing is being written to the CollectionSpace database and
the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're
taking your snapshot, otherwise you may find a few or many things that are
missing when it comes to recreate.  Finally, I would test that everything
is being backed up correctly by recreating your CollectionSpace instance in
a duplicate environment.

Best,

Yousuf

On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai mytsai2007@gmail.com wrote:

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace
for our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated
snapshots of our linode instance (daily and weekly), with an occasional
manual snapshot. We do not have a local backup (actually, I haven't even
looked into how that might be done).

Would be great to hear some best practices that you have adopted. And,
if you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012


Talk mailing list
Talk@lists.collectionspace.org
http://lists.collectionspace.org/mailman/listinfo/talk_lists
.collectionspace.org

--

Yousuf D. Nejati

Hi Ming, We don't directly deal with database backups any more, because we use a campus service. I'm attaching the script we used when we did, although it's not exactly what I'd use today, and it contains some things that are specific to our installation. Also see this issue for how we used to do the restore, but this may be outdated and/or depend on specifics of our installation: https://issues.collectionspace.org/browse/PAHMA-388 These docs have more information about filesystem backup vs. SQL dump: https://www.postgresql.org/docs/9.6/static/backup-file.html https://www.postgresql.org/docs/9.6/static/backup-dump.html Generally you want two backups: 1. A local backup -- in this case, local to Linode 2. A remote backup at a different geography/vendor There are three things you need to back up in a CollectionSpace system: a. Your configuration and customizations (aka source code) b. Your postgres databases c. Your uploaded data (nuxeo-server/data) A Linode backup/snapshot takes care of 1a, 1b, and 1c. You can use GitHub for 2a, but you don't have to. It's common because lots of us already use it, and it's good to have some kind of source control in any case. 2b and 2c will vary. You might use another cloud backup service. You might write some scripts that transfer the files to your own computer. Ray On Wed, Dec 6, 2017 at 1:27 AM, Ming Tsai <mytsai2007@gmail.com> wrote: > Hi Ray, > > Thanks for your input. Yes, totally okay with having CSpace off-line > during time of backup. Just checked my Linode and it gives me a choice of > 2-hour windows throughout the day where it'll do the daily backup. > > Would be happy to know more about both the 'SQL dump' versus the > 'filesystem level' approach. I'm somewhat comfortable with working the > command line though certainly no pro. > > Am I correct in assuming that most people backup to 3 locations (the cloud > service, github, and locally on their machine)? For me, it would be very > nice to have the instance backed-up on Linode, Github, and occasionally on > my local machine for peace of mind! > > Could you share some specific details/demo scripts on your approach? > > Thanks and best wishes, > Ming > > On Tue, Dec 5, 2017 at 10:25 PM, Ray Lee <rhlee@berkeley.edu> wrote: > >> Hi Ming, >> At Berkeley our backup situation is somewhat esoteric, due to the number >> of CollectionSpace instances we run, their size, and our use of services >> offered by campus IT. For most installations I think Yousuf's >> recommendations are spot on. >> >> We do SQL dump (pg_dump) for the database. The tradeoff with filesystem >> level backup is this: For filesystem level backup, the postgres server must >> be stopped. For pg_dump, the database can be running. But filesystem level >> backups can be restored just by restoring the filesystem, which is probably >> more or less a button click in linode. SQL dumps require some additional >> work on the command line to restore from the dump file. >> >> For me the relevant questions are: Does linode allow you to schedule when >> backups occur? Are you ok with having CSpace unusable around the time of >> the backup? >> >> Ray >> >> On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai <mytsai2007@gmail.com> wrote: >> >>> Hi Yousuf, >>> >>> Nice to hear from you! and thanks for getting back to me on the backup >>> question. >>> >>> I'm looking over some of my notes--it's been a long time since I've >>> looked at the backend of my instance--and I can't say I know which files >>> make up my 'source code.' Is it just the files in my .../tenants/ folders >>> in the 'application' and 'ui' layers? >>> >>> Would be great if you could tell me which folders contain my source code? >>> >>> Had a quick look at the PostgreSQL website and there are apparently 3 >>> methods of backup: SQL dump, file system level backup, and continuous >>> archiving. Which do you do? If you have a sample script that would be very >>> helpful. >>> >>> As for the daily snapshot, it is an automatic Linode backup service, so >>> I wasn't thinking of shutting down Cspace and the postgres database daily. >>> On my occasional backup, however, I could certainly shutdown both programs >>> before taking the snapshot. >>> >>> Thanks for any additional pointers. >>> >>> With best wishes, >>> Ming Tsai >>> >>> On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati <yousufnejati@gmail.com> >>> wrote: >>> >>>> Hi Ming, >>>> >>>> For backups my chosen methods are to (most of which you already know): >>>> >>>> 1) Storing the latest source code for my CollectionSpace instance up in >>>> remote git repository, i.e. Github, Bitbucket, Gitlab, etc. >>>> 2) Nightly PostgreSQL server database backups (you can create your own >>>> script or find some online) >>>> 3) Nightly snapshots of the host server (again, you can create your own >>>> script or find some online) >>>> >>>> Of course, if you're taking nightly snapshots of your host server, you >>>> should be capturing the servers entire state, which should allow you to >>>> recreate at any time. One thing I would suggest would be to shut down >>>> CollectionSpace and possibly your Postgres database (assuming it's only >>>> being used for CollectionSpace) before taking the snapshot. You want to >>>> make sure that nothing is being written to the CollectionSpace database and >>>> the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're >>>> taking your snapshot, otherwise you may find a few or many things that are >>>> missing when it comes to recreate. Finally, I would test that everything >>>> is being backed up correctly by recreating your CollectionSpace instance in >>>> a duplicate environment. >>>> >>>> Best, >>>> >>>> Yousuf >>>> >>>> On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai <mytsai2007@gmail.com> wrote: >>>> >>>>> Dear Talklist Subscribers, >>>>> >>>>> My name is Ming Tsai. We are a recent implementer of Collectionspace >>>>> for our family art and science foundation (tsaifoundation.org). >>>>> >>>>> With 60-70% of our inventory in Cspace now, I'd like to know how the >>>>> general Cspace community handles the BACKUP of their database and their >>>>> instance. >>>>> >>>>> Currently, our backup consists simply of having regular automated >>>>> snapshots of our linode instance (daily and weekly), with an occasional >>>>> manual snapshot. We do not have a local backup (actually, I haven't even >>>>> looked into how that might be done). >>>>> >>>>> Would be great to hear some best practices that you have adopted. And, >>>>> if you have specific methods, I'd be glad to see some procedures/steps. >>>>> >>>>> Thanks & best wishes, >>>>> Ming Tsai >>>>> >>>>> Treasurer & CTO >>>>> Tsai Art & Science Foundation >>>>> New York, NY 10012 >>>>> >>>>> _______________________________________________ >>>>> Talk mailing list >>>>> Talk@lists.collectionspace.org >>>>> http://lists.collectionspace.org/mailman/listinfo/talk_lists >>>>> .collectionspace.org >>>>> >>>>> >>>> >>>> >>>> -- >>>> >>>> *Yousuf D. Nejati* >>>> >>> >>> >>> _______________________________________________ >>> Talk mailing list >>> Talk@lists.collectionspace.org >>> http://lists.collectionspace.org/mailman/listinfo/talk_lists >>> .collectionspace.org >>> >>> >> >
MT
Ming Tsai
Fri, Dec 8, 2017 8:52 PM

Hi Ray,

Thanks for the breakdown of the tasks, the earlier sample bash script that
you used for the pgdump along with the links to the postgresql detailed
docs.

I will try to get a backup system off the ground using all this information
and let you know how it goes over the next couple weeks.

Thanks again and have a nice weekend.

Best wishes,
Ming

On Fri, Dec 8, 2017 at 12:49 AM, Ray Lee rhlee@berkeley.edu wrote:

Hi Ming,
We don't directly deal with database backups any more, because we use a
campus service. I'm attaching the script we used when we did, although it's
not exactly what I'd use today, and it contains some things that are
specific to our installation. Also see this issue for how we used to do the
restore, but this may be outdated and/or depend on specifics of our
installation: https://issues.collectionspace.org/browse/PAHMA-388

These docs have more information about filesystem backup vs. SQL dump:
https://www.postgresql.org/docs/9.6/static/backup-file.html
https://www.postgresql.org/docs/9.6/static/backup-dump.html

Generally you want two backups:

  1. A local backup -- in this case, local to Linode
  2. A remote backup at a different geography/vendor

There are three things you need to back up in a CollectionSpace system:
a. Your configuration and customizations (aka source code)
b. Your postgres databases
c. Your uploaded data (nuxeo-server/data)

A Linode backup/snapshot takes care of 1a, 1b, and 1c.

You can use GitHub for 2a, but you don't have to. It's common because lots
of us already use it, and it's good to have some kind of source control in
any case. 2b and 2c will vary. You might use another cloud backup service.
You might write some scripts that transfer the files to your own computer.

Ray

On Wed, Dec 6, 2017 at 1:27 AM, Ming Tsai mytsai2007@gmail.com wrote:

Hi Ray,

Thanks for your input. Yes, totally okay with having CSpace off-line
during time of backup. Just checked my Linode and it gives me a choice of
2-hour windows throughout the day where it'll do the daily backup.

Would be happy to know more about both the 'SQL dump' versus the
'filesystem level' approach. I'm somewhat comfortable with working the
command line though certainly no pro.

Am I correct in assuming that most people backup to 3 locations (the
cloud service, github, and locally on their machine)? For me, it would be
very nice to have the instance backed-up on Linode, Github, and
occasionally on my local machine for peace of mind!

Could you share some specific details/demo scripts on your approach?

Thanks and best wishes,
Ming

On Tue, Dec 5, 2017 at 10:25 PM, Ray Lee rhlee@berkeley.edu wrote:

Hi Ming,
At Berkeley our backup situation is somewhat esoteric, due to the number
of CollectionSpace instances we run, their size, and our use of services
offered by campus IT. For most installations I think Yousuf's
recommendations are spot on.

We do SQL dump (pg_dump) for the database. The tradeoff with filesystem
level backup is this: For filesystem level backup, the postgres server must
be stopped. For pg_dump, the database can be running. But filesystem level
backups can be restored just by restoring the filesystem, which is probably
more or less a button click in linode. SQL dumps require some additional
work on the command line to restore from the dump file.

For me the relevant questions are: Does linode allow you to schedule
when backups occur? Are you ok with having CSpace unusable around the time
of the backup?

Ray

On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai mytsai2007@gmail.com wrote:

Hi Yousuf,

Nice to hear from you! and thanks for getting back to me on the backup
question.

I'm looking over some of my notes--it's been a long time since I've
looked at the backend of my instance--and I can't say I know which files
make up my 'source code.' Is it just the files in my .../tenants/ folders
in the 'application' and 'ui' layers?

Would be great if you could tell me which folders contain my source
code?

Had a quick look at the PostgreSQL website and there are apparently 3
methods of backup: SQL dump, file system level backup, and continuous
archiving. Which do you do? If you have a sample script that would be very
helpful.

As for the daily snapshot, it is an automatic Linode backup service, so
I wasn't thinking of shutting down Cspace and the postgres database daily.
On my occasional backup, however, I could certainly shutdown both programs
before taking the snapshot.

Thanks for any additional pointers.

With best wishes,
Ming Tsai

On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati yousufnejati@gmail.com
wrote:

Hi Ming,

For backups my chosen methods are to (most of which you already know):

  1. Storing the latest source code for my CollectionSpace instance up
    in remote git repository, i.e. Github, Bitbucket, Gitlab, etc.
  2. Nightly PostgreSQL server database backups (you can create your own
    script or find some online)
  3. Nightly snapshots of the host server (again, you can create your
    own script or find some online)

Of course, if you're taking nightly snapshots of your host server, you
should be capturing the servers entire state, which should allow you to
recreate at any time.  One thing I would suggest would be to shut down
CollectionSpace and possibly your Postgres database (assuming it's only
being used for CollectionSpace) before taking the snapshot. You want to
make sure that nothing is being written to the CollectionSpace database and
the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're
taking your snapshot, otherwise you may find a few or many things that are
missing when it comes to recreate.  Finally, I would test that everything
is being backed up correctly by recreating your CollectionSpace instance in
a duplicate environment.

Best,

Yousuf

On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai mytsai2007@gmail.com
wrote:

Dear Talklist Subscribers,

My name is Ming Tsai. We are a recent implementer of Collectionspace
for our family art and science foundation (tsaifoundation.org).

With 60-70% of our inventory in Cspace now, I'd like to know how the
general Cspace community handles the BACKUP of their database and their
instance.

Currently, our backup consists simply of having regular automated
snapshots of our linode instance (daily and weekly), with an occasional
manual snapshot. We do not have a local backup (actually, I haven't even
looked into how that might be done).

Would be great to hear some best practices that you have adopted.
And, if you have specific methods, I'd be glad to see some procedures/steps.

Thanks & best wishes,
Ming Tsai

Treasurer & CTO
Tsai Art & Science Foundation
New York, NY 10012


Talk mailing list
Talk@lists.collectionspace.org
http://lists.collectionspace.org/mailman/listinfo/talk_lists
.collectionspace.org

--

Yousuf D. Nejati

Hi Ray, Thanks for the breakdown of the tasks, the earlier sample bash script that you used for the pgdump along with the links to the postgresql detailed docs. I will try to get a backup system off the ground using all this information and let you know how it goes over the next couple weeks. Thanks again and have a nice weekend. Best wishes, Ming On Fri, Dec 8, 2017 at 12:49 AM, Ray Lee <rhlee@berkeley.edu> wrote: > Hi Ming, > We don't directly deal with database backups any more, because we use a > campus service. I'm attaching the script we used when we did, although it's > not exactly what I'd use today, and it contains some things that are > specific to our installation. Also see this issue for how we used to do the > restore, but this may be outdated and/or depend on specifics of our > installation: https://issues.collectionspace.org/browse/PAHMA-388 > > These docs have more information about filesystem backup vs. SQL dump: > https://www.postgresql.org/docs/9.6/static/backup-file.html > https://www.postgresql.org/docs/9.6/static/backup-dump.html > > Generally you want two backups: > 1. A local backup -- in this case, local to Linode > 2. A remote backup at a different geography/vendor > > There are three things you need to back up in a CollectionSpace system: > a. Your configuration and customizations (aka source code) > b. Your postgres databases > c. Your uploaded data (nuxeo-server/data) > > A Linode backup/snapshot takes care of 1a, 1b, and 1c. > > You can use GitHub for 2a, but you don't have to. It's common because lots > of us already use it, and it's good to have some kind of source control in > any case. 2b and 2c will vary. You might use another cloud backup service. > You might write some scripts that transfer the files to your own computer. > > Ray > > On Wed, Dec 6, 2017 at 1:27 AM, Ming Tsai <mytsai2007@gmail.com> wrote: > >> Hi Ray, >> >> Thanks for your input. Yes, totally okay with having CSpace off-line >> during time of backup. Just checked my Linode and it gives me a choice of >> 2-hour windows throughout the day where it'll do the daily backup. >> >> Would be happy to know more about both the 'SQL dump' versus the >> 'filesystem level' approach. I'm somewhat comfortable with working the >> command line though certainly no pro. >> >> Am I correct in assuming that most people backup to 3 locations (the >> cloud service, github, and locally on their machine)? For me, it would be >> very nice to have the instance backed-up on Linode, Github, and >> occasionally on my local machine for peace of mind! >> >> Could you share some specific details/demo scripts on your approach? >> >> Thanks and best wishes, >> Ming >> >> On Tue, Dec 5, 2017 at 10:25 PM, Ray Lee <rhlee@berkeley.edu> wrote: >> >>> Hi Ming, >>> At Berkeley our backup situation is somewhat esoteric, due to the number >>> of CollectionSpace instances we run, their size, and our use of services >>> offered by campus IT. For most installations I think Yousuf's >>> recommendations are spot on. >>> >>> We do SQL dump (pg_dump) for the database. The tradeoff with filesystem >>> level backup is this: For filesystem level backup, the postgres server must >>> be stopped. For pg_dump, the database can be running. But filesystem level >>> backups can be restored just by restoring the filesystem, which is probably >>> more or less a button click in linode. SQL dumps require some additional >>> work on the command line to restore from the dump file. >>> >>> For me the relevant questions are: Does linode allow you to schedule >>> when backups occur? Are you ok with having CSpace unusable around the time >>> of the backup? >>> >>> Ray >>> >>> On Tue, Dec 5, 2017 at 6:55 AM, Ming Tsai <mytsai2007@gmail.com> wrote: >>> >>>> Hi Yousuf, >>>> >>>> Nice to hear from you! and thanks for getting back to me on the backup >>>> question. >>>> >>>> I'm looking over some of my notes--it's been a long time since I've >>>> looked at the backend of my instance--and I can't say I know which files >>>> make up my 'source code.' Is it just the files in my .../tenants/ folders >>>> in the 'application' and 'ui' layers? >>>> >>>> Would be great if you could tell me which folders contain my source >>>> code? >>>> >>>> Had a quick look at the PostgreSQL website and there are apparently 3 >>>> methods of backup: SQL dump, file system level backup, and continuous >>>> archiving. Which do you do? If you have a sample script that would be very >>>> helpful. >>>> >>>> As for the daily snapshot, it is an automatic Linode backup service, so >>>> I wasn't thinking of shutting down Cspace and the postgres database daily. >>>> On my occasional backup, however, I could certainly shutdown both programs >>>> before taking the snapshot. >>>> >>>> Thanks for any additional pointers. >>>> >>>> With best wishes, >>>> Ming Tsai >>>> >>>> On Mon, Dec 4, 2017 at 2:06 PM, Yousuf Nejati <yousufnejati@gmail.com> >>>> wrote: >>>> >>>>> Hi Ming, >>>>> >>>>> For backups my chosen methods are to (most of which you already know): >>>>> >>>>> 1) Storing the latest source code for my CollectionSpace instance up >>>>> in remote git repository, i.e. Github, Bitbucket, Gitlab, etc. >>>>> 2) Nightly PostgreSQL server database backups (you can create your own >>>>> script or find some online) >>>>> 3) Nightly snapshots of the host server (again, you can create your >>>>> own script or find some online) >>>>> >>>>> Of course, if you're taking nightly snapshots of your host server, you >>>>> should be capturing the servers entire state, which should allow you to >>>>> recreate at any time. One thing I would suggest would be to shut down >>>>> CollectionSpace and possibly your Postgres database (assuming it's only >>>>> being used for CollectionSpace) before taking the snapshot. You want to >>>>> make sure that nothing is being written to the CollectionSpace database and >>>>> the CollectionSpace blobs (i.e. nuxeo-data/data) directory while you're >>>>> taking your snapshot, otherwise you may find a few or many things that are >>>>> missing when it comes to recreate. Finally, I would test that everything >>>>> is being backed up correctly by recreating your CollectionSpace instance in >>>>> a duplicate environment. >>>>> >>>>> Best, >>>>> >>>>> Yousuf >>>>> >>>>> On Mon, Dec 4, 2017 at 3:08 AM, Ming Tsai <mytsai2007@gmail.com> >>>>> wrote: >>>>> >>>>>> Dear Talklist Subscribers, >>>>>> >>>>>> My name is Ming Tsai. We are a recent implementer of Collectionspace >>>>>> for our family art and science foundation (tsaifoundation.org). >>>>>> >>>>>> With 60-70% of our inventory in Cspace now, I'd like to know how the >>>>>> general Cspace community handles the BACKUP of their database and their >>>>>> instance. >>>>>> >>>>>> Currently, our backup consists simply of having regular automated >>>>>> snapshots of our linode instance (daily and weekly), with an occasional >>>>>> manual snapshot. We do not have a local backup (actually, I haven't even >>>>>> looked into how that might be done). >>>>>> >>>>>> Would be great to hear some best practices that you have adopted. >>>>>> And, if you have specific methods, I'd be glad to see some procedures/steps. >>>>>> >>>>>> Thanks & best wishes, >>>>>> Ming Tsai >>>>>> >>>>>> Treasurer & CTO >>>>>> Tsai Art & Science Foundation >>>>>> New York, NY 10012 >>>>>> >>>>>> _______________________________________________ >>>>>> Talk mailing list >>>>>> Talk@lists.collectionspace.org >>>>>> http://lists.collectionspace.org/mailman/listinfo/talk_lists >>>>>> .collectionspace.org >>>>>> >>>>>> >>>>> >>>>> >>>>> -- >>>>> >>>>> *Yousuf D. Nejati* >>>>> >>>> >>>> >>>> _______________________________________________ >>>> Talk mailing list >>>> Talk@lists.collectionspace.org >>>> http://lists.collectionspace.org/mailman/listinfo/talk_lists >>>> .collectionspace.org >>>> >>>> >>> >> >