Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Thank you very much, i anticipated that, but as the cluster is in production hoped to avoid it :) AFAIK it is possible to upgrade servers one-by-one while retaining protocol compatibility (eg update to latest 8.4.x) , am i right? On 04.07.2014 16:25, Lars Ellenberg wrote: > On Mon, Jun 30, 2014 at 10:35:38AM +0400, Igor Novgorodov wrote: >> Hello everyone, i need some help. >> >> I've got 2 servers in DRBD mirror, replication link is 4x1Gbit in >> balance-rr bond, resulting in about 3.8 Gbit/s as measured by iperf. >> No errors on interfaces at all, the replication never fails under >> normal conditions, it's been up for almost a year. >> >> But when i decide to do a verification of a DRBD device, it goes >> fine for some time (each time different), but then it fails. >> Kernel is 3.4.42, DRBD is built as external module version 8.4.3 > Upgrade your DRBD module. > > While this is a typical standard answer, > it will in fact help. > > What happened is what I call a distributed resource starvation deadlock, > and it has been fixed since. > >> Any help would be appreciated. >> >> dmesg: >> [129075.674385] block drbd0: conn( Connected -> VerifyS ) >> [129075.674416] block drbd0: Starting Online Verify from sector 44054936 >> [132983.906214] d-con VM_STORAGE2_1: [drbd_w_VM_STORA/3871] >> sock_sendmsg time expired, ko = 6