Decide what LVs to host on lizard rotating drives
We have created a dedicated
spinninglizard VG on lizard's rotating drives when we installed new SSDs. It's empty at the moment, so we have to decide which LVs are going to be on this VG and then move them.
#4 Updated by bertagaz over 1 year ago
- Status changed from Confirmed to In Progress
- Assignee changed from bertagaz to intrigeri
- % Done changed from 0 to 20
- QA Check set to Info Needed
I guess it'll be a subset of what we had there previously (IIRC I had put quite some thought into it).
jenkins-data LV is 300G big. We have planned to grow the
isos LVs to 200G and 240G respectively (which include both occurences of each, the gitolite and git-annex clone). So my proposal is to move out the
jenkins-data one, and put all
torbrowser-archive LVs on this rotating drive, as well as the rsync-data one. This way we should be close to the disk space limit (which is 558g) and still have a bit of free space just in case. What's your opinion?
#5 Updated by intrigeri over 1 year ago
- Assignee changed from intrigeri to bertagaz
- QA Check changed from Info Needed to Dev Needed
jenkins-dataLV is 300G big. We have planned to grow the
isosLVs to 200G and 240G respectively (which include both occurences of each, the gitolite and git-annex clone). So my proposal is to move out the
jenkins-dataone, and put all
torbrowser-archiveLVs on this rotating drive, as well as the rsync-data one.
ACK wrt. moving back the
jenkins-data one and
rsync-data to rotating drives.
torbrowser-archive ones, I would move only those that are rarely used i.e. the ones that are assigned to
puppet-git. And the ones that are used during each ISO build (assigned to
www) could remain on faster SSDs.
So essentially we would get back to exactly what we had previously.
This way we should be close to the disk space limit (which is 558g) and still have a bit of free space just in case. What's your opinion?
It sounds like you're trying to move as much stuff as possible to slower, rotating drives. I don't understand why we would want to do that: it feels wrong that adding SSDs results in degraded I/O performance, no? If you have arguments in favour of this change, please state them before asking me for my opinion (i.e. implicitly asking me to do the reasoning myself, as I've already mentioned in a meta comment a couple months ago). Thanks in advance!