Customer schedules a weekly PACI backup, but it is not created for a long time
No free space on backup nodes, or misconfiguration in PACI DB
Backups on PACI IM run according to settings in
<!-- backup job details --> <backups batchSize="10" intervalInMin="1" maxDurationInMin="200"/>
by default PACI runs backups in 10 VEs batches. It means that free space on Backup node is checked using the sum of 10 VEs that should be backed up.
In this case try to calculate maximum size of the 10 biggest VEs:
im=# with topdisks as (select size from ve_storage where ve_ref in (select id from ve where template_id in (select id from templates where ((os_type in (select type from os_types where family='linux') and technology='CT')) or technology='VM')) order by size desc limit 10) select sum(size) from topdisks; sum ----- 1625
In this example for performing 10 biggest VEs backup batch - 1.65 T of free space is required. Please verify that Backup nodes have enough space.
As a workaround, you may change batchSize to "5" for example - in this case backup task will ran smaller batches that will require less space. Please note that any changes in
IM-config.xml require PACI-IM service restart.
Another kind of cause is misconfiguration in PACI DB:
backups_to_removetable - it shouldn't have many backups, if it is - indication of some problems with backups removal. In this case, PACI IM logs should be checked.
active_backupstable - if you will notice any long running backups (more than 1-2 days) - indication of terracotta or dispatcher failures.
to fix it - remove long running backups:
im=# select id,ve_ref,started,ended,stage from active_backups; id | ve_ref | started | ended | stage ------+--------+-------------------------------+-------+--------- 6061 | 381 | 2013-08-02 12:00:01.276774+02 | | STARTED 6084 | 398 | 2013-08-03 06:02:03.397072+02 | | STARTED (2 rows) im=# delete from active_backups; DELETE 2
The info about running backups will be re-read on the next backup task run.