Let me recap what we did last weekend.
* The Coin storage now has more storage space available. This should mean
less rebooting the system for cleanup.
* We also noticed a VMware teak that increases the I/O from the storage.
This could potentially make Coin more stable as the bandwidth doesn’t run out.
* We now have LVM installed, so further expansion of the storage is easier
Because we installed LVM, we had to redo the partitions. This required us to
move data about a lot. We could have just deleted the logs and artifacts and
let Coin rebuild them (we would have gotten the system up already on Friday
evening), but by copying over all the data from a backup we enabled Coin to
resume like nothing had happened (didn’t need to rebuild all the existing
artifacts). It also enabled all the current logs to be visible for our company
internal users that read the data straight from Coin instead of
testresults.qt.io where they are published. With all the flaky autotests we
had, we might have ended up trying to recover from lost artifacts until
Saturday morning anyway. So the delay wasn’t bad if we think about it that way 😊
On an unrelated note. We’ve reverted some of the build hosts to Ubuntu 16.04 as
the underlying platform. It looks like it’s just as stable if not more stable
than the 17.xx ones. At least as of yet, not a single one has crashed! If this
continues, this is _very_ good news for the CI. It will knock down a very
annoying problem where the hosts that’s running all the virtual machines kept
crashing and failing all the builds. Let’s hope this trend continues.
From: Simon Hausmann
Sent: perjantai 16. helmikuuta 2018 17.10
To: Tony Sarajärvi <tony.saraja...@qt.io>; firstname.lastname@example.org
Subject: Re: CI going down for update
Thanks for the update! Good luck and thanks for working on this :)
From: Tony Sarajärvi
Sent: Friday, February 16, 2018 3:44:49 PM
Subject: RE: CI going down for update
We have to do things the hard way, meaning copying over a lot of stuff instead
of just converting stuff. No unexpected things have poped up during the hard
way however, but the copy phase takes a lot of time. ETA (or nothing really
arrives now… so EToC): in a few hours.
From: Tony Sarajärvi
Sent: perjantai 16. helmikuuta 2018 9.20
Subject: CI going down for update
We’ll take the CI down for update for a short while. The web page won’t work at
all during that time, since we’ll close the whole server.
Thanks for your patience.
Development mailing list