Issue with Restore Jobs Over 300GB in Bacularis

thcjusz

You are right, but on server like samba it's hard. To many new folders are adding. So everytime when new folder is created i needed to add this in fileset. Referring back to the post about bvfs
*.bvfs_lsdirs jobid=9145 path=
9785 0 0 0 A A A A A A A A A A A A A A .
9784 0 0 0 A A A A A A A A A A A A A A /
*.bvfs_clear_cache yes
OK
*.bvfs_update jobid=9145
*.bvfs_lsdirs jobid=9145 path=
9785 0 0 0 A A A A A A A A A A A A A A .
9784 0 0 0 A A A A A A A A A A A A A A /
idk why is still 0

thcjusz

i will send u a screen, after .bvfs_update still got a problem

gani

Hi @thcjusz,

Your output is fully correct Bvfs output:

*.bvfs_lsdirs jobid=9145 path=
9785 0 0 0 A A A A A A A A A A A A A A .
9784 0 0 0 A A A A A A A A A A A A A A /

Now you can browse through the directory three:

.bvfs_lsdirs jobid=9145 path=/

and so on.

If in the second wizard step the HTTP request still times out, this is not not a matter of Bvfs cache.

The attached screenshot it very small and this is difficult to see what is on it. If I am not mistaken and I am seeing correctly this is /monitor/ path in that request, right? The monitor is not used during restore. Could you provide us this information what HTTP request exactly times out and with what HTTP error codes? These information are in the web browser developer console (network tab). They can be also available in the web server access log.

Thanks.

Best regards,
Marcin Haba (gani)

gani

One more note. This command:

.bvfs_clear_cache yes

clears all Bvfs cache. If the cache was collected for this big job in Runscript, then executing the above command cleared it for all jobs and it needs to be built again. You can use Bconsole .bvfs_update jobid=xxx,yyy,zzz for that.

Worth to know is that the Bvfs cache can be built incrementaly for selected jobids, but for clearing it is cleared for all jobs. There does not exist any command to clear cache for single jobids.

Best regards,
Marcin Haba (gani)

thcjusz

i can't copy, after doing prscr image is small, i will write

thcjusz

gani

Hi @thcjusz,

thcjusz PRADO Ajax callback error 0 (timeout)
errorHandler http://192.168.186.29:9095/assets/8498804c/activecontrols/ajax3.js:356

It looks that you do not use Bacularis, but Baculum. This is different web interface from Bacularis. I encourage to install Bacularis 🙂 It is much more modern and dynamically developed.

For your Bacula and restore, you can try the following two commands in Bconsole:

.bvfs_get_jobids jobid=BACKUP_JOBID_TO_RESTORE
XXXXX,YYYYY,ZZZZZ

.bvfs_update jobid=XXXXX,YYYYY,ZZZZZ

Where:

BACKUP_JOBID_TO_RESTORE - jobid to restore
XXXXX,YYYYY,ZZZZZ - output jobids from .bvfs_get_jobids

For example for jobid 56112:

.bvfs_get_jobids jobid=56112
56003,56081,56112

.bvfs_update jobid=56003,56081,56112

And then on the web interface go to this jobid BACKUP_JOBID_TO_RESTORE and try restore.

Thanks in advance for sharing us your feedback.

Best regards,
Marcin Haba (gani)

thcjusz

after trying to restore, the Bacularis freeze, GUI won't work,. I need to kill postgres process and restart httpd.service to get to work again

thcjusz

Ty for answer Marcin. I have bacularis installed but in production environment, i migrated servers from old baculum to bacularis. I prefer doing test on baculum to avoid problems environment on new And i have problem in old like in new one. The amount of files dont have difrience, when i try restore incremental of 100 files from this job it freezez and get a error, other 26 jobs on difrient servers working great with restore job(100 of 200000 files). but only this one.

gani

Hi @thcjusz,

thcjusz I have bacularis installed but in production environment

OK, great.

thcjusz The amount of files dont have difrience, when i try restore incremental of 100 files from this job it freezez and get a error,

This is very interesting. I would propose to check the system and hardware.

thcjusz i will do what You wrote and will let You know

Yes, thanks. The most important is to check how long the .bvfs_update command will be executed and what takes the most resources (CPU, I/O, memory ...) in the system and if there is something strange or unexpected in these results. We know that PostgreSQL takes it, but it does not have to mean that this is the bottleneck because the source can be lower than database.

Best regards,
Marcin Haba (gani)

thcjusz

i will do what You wrote and will let You know

thcjusz

Hi Marcin,
On bacualris i go error after backup.
Runscript: AfterJob returned non-zero status=1. ERR=Operacja niedozwolona
TELBBACULA1R-dir JobId 509: console command: run AfterJob "".bvfs_update jobid=509""
The bvfs_update working for 4,5h on baculum. I noticed that in postgresql.conf i have def configuration.
maybe u know how to solve it. Did bfvs lunched?
On baculum still .bvfs_updtae for one job still working:

- Memory -

shared_buffers = 128MB # min 128kB

(change requires restart)

#huge_pages = try # on, off, or try

(change requires restart)

#temp_buffers = 8MB # min 800kB
#max_prepared_transactions = 0 # zero disables the feature

(change requires restart)

Caution: it is not advisable to set max_prepared_transactions nonzero unless

you actively intend to use prepared transactions.

#work_mem = 4MB # min 64kB
#hash_mem_multiplier = 1.0 # 1-1000.0 multiplier on hash table work_mem
#maintenance_work_mem = 64MB # min 1MB
#autovacuum_work_mem = -1 # min 1MB, or -1 to use maintenance_work_mem
#logical_decoding_work_mem = 64MB # min 64kB
#max_stack_depth = 2MB # min 100kB
#shared_memory_type = mmap # the default is the first option
this can be the issue of low bfvs_update.

thcjusz

TELBBACULA1R-dir JobId 509: console command: run AfterJob "".bvfs_update jobid=509""
maybe when i put command in runscript>console it shpould be like .bvfs_update jobid=509 without "

thcjusz

ok bvfs_update end running, and after that i can restore via GUI, Thank You Marcin. but could u help me with that error?

gani

Hi @thcjusz,

thcjusz ok bvfs_update end running, and after that i can restore via GUI, Thank You Marcin.

Great news and good job 👍️ I am glad that now you are able to restore 🙂

For the proper database configuration, I don't know much about it. The default database settings after db installation is usually not optimal because it is generic and it is not adapted to your server resources and use case. This is why the database needs tuning. In the Bacularis documentation you can find useful tips where to start with it:

https://bacularis.app/doc/brief/optimization.html#database

Maybe on this Bacularis Group is any database administrator that could help with this settings? I encourage for that.

For the quotes in ".bvfs_update jobid=509", yes, your notice is fully correct. They need to be removed to make it working.

Best regards,
Marcin Haba (gani)

thcjusz

Ty Marcin,
Without " script working like a charm 🙂 but i need ,bvfs_update of full backup samba, i see difrience on bacularis when .bfvs_update working. Tune DB helped. But 3,5 milion files it take a lot of time i guess.

gani

Hi @thcjusz,

thcjusz Without " script working like a charm 🙂 but i need ,bvfs_update of full backup samba, i see difrience on bacularis when .bfvs_update working. Tune DB helped. But 3,5 milion files it take a lot of time i guess.

Excellent 🙂

With 3.5 milion of files in many cases it is worth to consider using virtual full backup or progressive virtual full. This way Bacula does the big full backup only once, later it adds incrementa and/or differential changes.

Once you already have files from this big backup in Bvfs cache, you need to be careful to not run clearing this Bvfs cache because then all cache will need to be created from scratch once again. For milions of records, recreating them can be painful 🙂

Best regards,
Marcin Haba (gani)

thcjusz

Hi Marcin,
today i got weird message 31-mar 11:05 TELBBACULA1R-dir JobId 0: .bvfs_update jobid: is an invalid command. I got it twice today. Yes Virtual i good option but how to do it if i got already normal full? and when i try to set virtual in existing jobl i got error?

thcjusz

nvm Marcin about this error log i solved it 🙂 but how to do virtual or maybe how to add virtual to existing job without any error? Does it need create new Volume?

« Previous Page Next Page »