Getting time scanning logs on specific hour delay the backup...

[baph0m3t]

Hi

I’m doing backup each 30min,… and all my process is more less 21-23min until Monday.

xtrabackup_normal_job.PNG

Since yesterday and today I’m starting to get backup from near to my limit 30min more less each 2 hours… what could be happen?

so reading in my log of xtrabackup job I found scanning log time is more longer than the rest intermediates jobs

for example:

00.00 is normal get almost 27 because is full but…the next incrementals until each 2 hour are.

00.30 19min (scanning logs 3min)

01.00 20min (scanning logs 4min)

01.30 20min (scanning logs 4min)

02.00 27min (scanning logs 9min)

02.30 19min (scanning logs 3min)

03.00 20min (scanning logs 4min)

03.30 20min (scanning logs 4min)

04.00 27min (scanning logs 9min)

xtrabackup_delay_job_2h_scanning.PNG

and so on…to get today this

xtrabackup_delay_job_2h.PNG

The script of backup is working fine since 2 month ago… could be some kind of process to save in the logs to can do this kind of delay when hit this coincidence hours because was registred into binlogs ?

[matthewb]

@baph0m3t Taking a backup every 30m is quite excessive. You should consider setting up a replica and use the replica as your “hot standby”.

What do the sizes of your tables look like? Are they growing in size? The larger your tables, the longer the backup will take.

Also, the backup process itself is extremely disk IO intensive. If you have other things happening on the system at the same time, the apps will fight for disk IO. Watch your await% in top while the backup happens. If it goes above 2-5%, then you have disk issues.