Bug #8414
closedTeuthology: getaddrinfo vpmXXX.front.sepia.ceph.com: Name or service not known
0%
Description
I have been getting this message after teuthology-lock'ing a vpm when there are very few vpms available. This appears to be a socket.getaddrinfo issue deep within the bowels of python. 5715 mentions this problem but did not deal with it.
I have seen this happen for several different vm's. I would unlock a vm, and lock another one and get this same problem. I also only see this problem when we are in the state where there are few vps available (scheduled_teuthology@teuthology has 92 vms).
Since this could be at the socket level, this problem may still be around even if we adopt a new locking scheme.
Updated by Anonymous almost 10 years ago
Okay. I might have messed up here. I think that the vms that I am grabbing are not kosher. Downburst, when run on them, raises the error "NameError: Image not found on server at http://ceph.com/cloudinit/." So I suspect that there is something wrong with the vms that I am grabbing. The scheduled_teuthology runs are consuming 90 vms, and it's quite possible that what's left are the bad ones.
Updated by Anonymous almost 10 years ago
- Status changed from New to Closed
I have not seen this recently. Given the recent changes to teuthology, this may not be an issue anymnore.