Hello,
following an upgrade of the OS on our Bareos server (Debian 8 -> 11), we
upgraded Bareos to version 22.0.2 from 17.x. using the community packages
from the site; the upgrade went well and system is running correctly from
the old configuration, needing only minor changes to disable the TLS
support.
Two days ago i had to restore a file from the last full backup of our
fileserver, the restore went well but took around 7 hours to complete (for a
single file); with the old version the time spent was always around 2-3
minutes average. Our Bareos installation saves data on file volumes, not
tapes, and doing some tests we verified that the restore took 7 hours
because it started reading the volume from the beginning of the job until it
found the file requested; the size of full job is around 5.5 TB in a single
Volume and the file was near the end, so Bareos had to do almost a full read
of the volume before finding it.
Digging a little bit further we found that the old version, during a backup,
wrote a sort of "index" in the jobmedia table, so it can seek near the
position of the requested file and make the restore a lot faster, the backup
from the version 22 has only one row indicating the starting and ending
position of the job in the volume:
[backup done with the old version of Bareos, full job on a single volume,
around 19 GB]:
+--------+-------+----------+----------------+---------------------+---------------+
| jobid | level | jobfiles | jobbytes | starttime |
volumename |
+--------+-------+----------+----------------+---------------------+---------------+
| 148366 | F | 53,780 | 19,403,934,091 | 2023-01-02 04:35:38 |
win-infopub-1 |
bareos=# select * from jobmedia where jobid=148366;
jobmediaid | jobid | mediaid | firstindex | lastindex | startfile |
endfile | startblock | endblock | volindex | jobbytes
------------+--------+---------+------------+-----------+-----------+---------+------------+------------+----------+----------
8079653 | 148366 | 452 | 46067 | 53780 | 5 |
5 | 2845841927 | 3277427288 | 20 | 0
8079652 | 148366 | 452 | 30131 | 46067 | 5 |
5 | 1845905960 | 2845841926 | 19 | 0
8079645 | 148366 | 452 | 14749 | 30131 | 5 |
5 | 845970039 | 1845905959 | 18 | 0
8079640 | 148366 | 452 | 7721 | 14749 | 4 |
5 | 4141001410 | 845970038 | 17 | 0
8079636 | 148366 | 452 | 7512 | 7721 | 4 |
4 | 3141065414 | 4141001409 | 16 | 0
8079634 | 148366 | 452 | 7512 | 7512 | 4 |
4 | 2141129415 | 3141065413 | 15 | 0
8079633 | 148366 | 452 | 7473 | 7512 | 4 |
4 | 1141193420 | 2141129414 | 14 | 0
8079631 | 148366 | 452 | 7407 | 7473 | 4 |
4 | 141257426 | 1141193419 | 13 | 0
8079630 | 148366 | 452 | 7369 | 7407 | 3 |
4 | 3436288733 | 141257425 | 12 | 0
8079629 | 148366 | 452 | 7328 | 7369 | 3 |
3 | 2436352734 | 3436288732 | 11 | 0
8079627 | 148366 | 452 | 7298 | 7328 | 3 |
3 | 1436416734 | 2436352733 | 10 | 0
8079626 | 148366 | 452 | 7274 | 7298 | 3 |
3 | 436480745 | 1436416733 | 9 | 0
8079624 | 148366 | 452 | 6226 | 7274 | 2 |
3 | 3731512054 | 436480744 | 8 | 0
8079620 | 148366 | 452 | 6058 | 6226 | 2 |
2 | 2731576066 | 3731512053 | 7 | 0
8079617 | 148366 | 452 | 3325 | 6058 | 2 |
2 | 1731640078 | 2731576065 | 6 | 0
8079613 | 148366 | 452 | 3317 | 3325 | 2 |
2 | 731704078 | 1731640077 | 5 | 0
8079607 | 148366 | 452 | 3317 | 3317 | 1 |
2 | 4026735374 | 731704077 | 4 | 0
8079598 | 148366 | 452 | 1789 | 3317 | 1 |
1 | 3026799387 | 4026735373 | 3 | 0
8079590 | 148366 | 452 | 244 | 1789 | 1 |
1 | 2026863387 | 3026799386 | 2 | 0
8079582 | 148366 | 452 | 1 | 244 | 1 |
1 | 1026927388 | 2026863386 | 1 | 0
[backup done with version 22, full job on a single volume, around 5,5 TB]:
+--------+-------+-----------+-------------------+---------------------+---------------+
| jobid | level | jobfiles | jobbytes | starttime |
volumename |
+--------+-------+-----------+-------------------+---------------------+---------------+
| 149513 | F | 2,886,927 | 5,637,588,281,412 | 2023-01-21 00:30:02 |
fileserver-12 |
bareos=# select * from jobmedia where jobid=149513;
jobmediaid | jobid | mediaid | firstindex | lastindex | startfile |
endfile | startblock | endblock | volindex | jobbytes
------------+--------+---------+------------+-----------+-----------+---------+------------+-----------+----------+----------
8086618 | 149513 | 2146 | 1 | 2886927 | 6 | 1320 | 1489561298 | 465483062 |
1 | 0
(the jobs are from 2 different clients because the first job was one of the
few done with the old version remaining in the catalog)
There's someting that has to be changed in the configuration to make it work
as before? I tried to search the documentation but cannot find anything
about it.. I tried to activate the checkpoint function but i cannot see any
visible difference regarding this behaviour.
Thank you
Daniele
--
You received this message because you are subscribed to the Google Groups
"bareos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/bareos-users/WC20230126122232.490768%40zensistemi.com.
