19 octobre 2016 09:21 "Art H." <art@xxxxxxxxxxxxxx> a écrit:
Also, this morning I've set up a first bot, provided by Arfonzo. It'sover the current package list.
currently (slowly) running
Ahoy friends,
I've powered off the buildslave VM this morning to move it onto a less used
physical disk. The disk
I/Os were curiously high on the Haiku VM last night.
* From my tests it does seem to have improved the disk I/O.
* Not sure why, but I feel there's some kind of issue with Haiku, Hyper-V and
SSHDs (it's not
sharing nicely with other VMs: I run a lot of VMs on shared physical disks,
nothing Linux, Windows
or otherwise slows down no matter how heavy the usage, like we saw yesterday
in the Haiku VM)
* I could not see how to re-start the buildslave process after reboot, sorry.
Pulkomandy: Can you please re-start the buildslave when you have a chance?
Instructions/otherwise
would be handy for us buildslave hosters, in the future, if manual steps are
required from our end.
Also, I noticed you re-tried from master 10 times in quick succession. I
think it should just be
infinite, or 1000 (if it's going to be such short retries). When the
buildslave VMs go down here,
such as the VM host restarting, it could be offline for up to 30 minutes as
the entire
infrastructure of VMs reboot (suspends up to 20 VMs, reboot host, restore 20
VMs--that takes a
while).