[PVE-User] pve-zsync processes

[Mark Adams]

Hi All,

I've been using pve-zsync for a few months - it seems to work pretty well.

However, I have just noticed it doesn't seem to be terminating itself
correctly. at present I have around 800 pve-zsync processes (sleeping)
which all seems to be duplicates. (I would expect 1 per VMID?)

Has anyone noticed this behaviour? any idea why or how to stop it?

Best Regards,
Mark

[Wolfgang Link]

Hi,

this indicates that the sync time is to low.
cron fork every (default) 15 minutes a pve-zsync process.
If the former pve-zsync process is not finished, it will wait until the former process is done.

You should rise your sync interval this can be done in the /etc/cron.d/pve-zsync.

Best Regards,

Wolfgang Link

[Mark Adams]

Hi Wolfgang,

So does this mean that all those processes are sitting in a "queue" waiting
to execute? wouldn't it be more sensible for the script to terminate if a
process is already running for the same job?

Regards,
Mark

Post by Wolfgang Link
Hi,
this indicates that the sync time is to low.
cron fork every (default) 15 minutes a pve-zsync process.
If the former pve-zsync process is not finished, it will wait until the
former process is done.
You should rise your sync interval this can be done in the
/etc/cron.d/pve-zsync.
Best Regards,
Wolfgang Link

[Wolfgang Link]

Post by Mark Adams
So does this mean that all those processes are sitting in a "queue" waiting
to execute? wouldn't it be more sensible for the script to terminate if a
process is already running for the same job?

No because as I wrote 15 is default, but we have many user which have longer intervals like 1 day.
If you would quit the process you would skip one day.

[Thomas Lamprecht]

Post by Wolfgang Link

Post by Mark Adams
So does this mean that all those processes are sitting in a "queue" waiting
to execute? wouldn't it be more sensible for the script to terminate if a
process is already running for the same job?

No because as I wrote 15 is default, but we have many user which have longer intervals like 1 day.
If you would quit the process you would skip one day.

But that could simply be solved that only one process may be queued,
i.e., if one is active a the next starts the it waits. If the next(-next)
start and the former still waits then it just exists immediately.
With this we'd never miss a sync cycle and have a maximal queue of 1,
not 500 or more...