On 11/5/19 1:21 PM, Thomas Lamprecht wrote:
On 11/5/19 1:04 PM, Alexandre DERUMIER wrote:
Hi,
be carefull of timeouts, because it could hang pvestatd.
(That's why it was implemented in udp for graphite && influxdb)
That's true, as almost always there's a reason why seemingly simple
things are not already done.. So we need to be a bit more intelligent here.
Run the update as fork (with timeout?) and if we hit the timeout we
skip updating this specific instance for a bit and log a sane error,
as an idea.
of course, you're both right, thanks for catching this..
run with timeout could work for now, but it seems that
sooner or later, we have to get serious about that statd
rework...
_______________________________________________
pve-devel mailing list
pve-devel@pve.proxmox.com
https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
