Re: [OpenAFS] broken callbacks

2016-04-22 Thread Andreas Hirczy
Mark Vitale writes: > nFEs is the number of FileEntry table entries currently in use by the > fileserver. > One entry is used for each unique file (vnode) the fileserver has given > callbacks for. > > nCBs is the number of CallBack table entries currently in use by the

Re: [OpenAFS] broken callbacks

2016-04-21 Thread Benjamin Kaduk
On Thu, 21 Apr 2016, Andreas Hirczy wrote: > Benjamin Kaduk writes: > > > My draft change to auto-tune fileserver parameters for a target memory > > footprint at https://gerrit.openafs.org/#/c/11629/ just uses all the > > remaining memory footprint after the other parameters are

Re: [OpenAFS] broken callbacks

2016-04-21 Thread Andreas Hirczy
Benjamin Kaduk writes: > On Wed, 20 Apr 2016, Andreas Hirczy wrote: > >> >>I'd rather prefer not to increase -cb, since this seems to be not our >>usual usage pattern. > > Increasing -cb is pretty cheap; IIRC it's about 32 bytes of memory per > callback, and the runtime

Re: [OpenAFS] broken callbacks

2016-04-20 Thread Benjamin Kaduk
On Wed, 20 Apr 2016, Andreas Hirczy wrote: > >I'd rather prefer not to increase -cb, since this seems to be not our >usual usage pattern. Increasing -cb is pretty cheap; IIRC it's about 32 bytes of memory per callback, and the runtime overhead only comes into play at shutdown. My draft

Re: [OpenAFS] broken callbacks

2016-04-20 Thread Mark Vitale
On Apr 20, 2016, at 11:54 AM, Andreas Hirczy wrote: > In that context 3 questions regarding nFEs and nCBs entered my mind: > > - What's the meaning of those? nFEs is the number of FileEntry table entries currently in use by the fileserver. One entry is used for each unique

[OpenAFS] broken callbacks

2016-04-20 Thread Andreas Hirczy
Hi! To hunt down some performance problems on our AFS servers I started to integrate some metrics to our Ganglia monitoring system via a dirty hack; see for example . In that context 3