What did you change, “about a month ago” ? I have TCP problems when a node is DOWN, wherein amchecking it will cause problems with several other nodes (due to TCP timeout values) BUT your nodes are up.
“Broken pipe” plus no logs on the client sounds like the system level (outside of amanda?) TCP is having troubles; the client is never getting an amanda started at all. But I can’t think why that would happen. Deb Baddorf > On Jul 27, 2016, at 1:20 PM, Chris Nighswonger <[email protected]> > wrote: > > No takers? > > Additionally, on the failing client side an amcheck log is created > *only* if amcheck is run against that client. When amcheck is run > against the entire job (all clients) no amcheck log is created on the > clients that fail. > > > > On Tue, Jul 26, 2016 at 9:09 AM, Chris Nighswonger > <[email protected]> wrote: >> tcpdump shows that there is no (zero) data flow between the server and >> failed clients when running amcheck on multiple clients at once. Data >> does flow when checking a single client. >> >> This only started about a month ago. Maybe this is a bug? >> >> On Mon, Jul 25, 2016 at 4:16 PM, Chris Nighswonger >> <[email protected]> wrote: >>> Here is some snips of output illustrating the problem: >>> >>> >>> backup@scriptor: amcheck -c campus scriptor >>> >>> Amanda Backup Client Hosts Check >>> -------------------------------- >>> Client check: 1 host checked in 2.372 seconds. 0 problems found. >>> >>> (brought to you by Amanda 3.3.6) >>> backup@scriptor: amcheck -c campus masada >>> >>> Amanda Backup Client Hosts Check >>> -------------------------------- >>> Client check: 1 host checked in 2.074 seconds. 0 problems found. >>> >>> (brought to you by Amanda 3.3.6) >>> backup@scriptor: amcheck -c campus >>> >>> Amanda Backup Client Hosts Check >>> -------------------------------- >>> WARNING: scriptor: selfcheck request failed: error sending REQ: write >>> error to: Broken pipe >>> WARNING: masada: selfcheck request failed: error sending REQ: write >>> error to: Broken pipe >>> >>> >>> Server version is 3.3.6 >>> Client version on scriptor is 3.3.6 and on masada is 3.3.3 >>> >>> When the backup runs over night, the same error appears on these clients. >>> >>> Any thoughts on what might be going on here? >>> >>> Kind regards, >>> Chris
