That's the singular major problem with using collectl to parse command output. Just when you think you have something working, they go an change it out from under you. So clearly this is case where I screwed up but lucked out. Thanks for catching it and saving a lot of people a lot of headaches. It is in the new version and I'm just waiting on a few people to get back to me who've been doing some testing. Would you like a copy as well or do you just want to wait another week or so? -mark
On Fri, Mar 8, 2013 at 4:23 PM, Joe Digilio <[email protected]> wrote: > I was surprised that nobody else had reported it, until I dug a little > deeper. > > The problem started after I applied a bunch of updates to my RHEL5 systems, > including upgrading infiniband-diags from 1.5.3-1.el5 to 1.5.12-2.el5. > Indeed the output format of perfquery has changed. I believe 1.5.3 did not > have "(CapMask: 0x00)". With collectl 3.6.0 (the version I'm running) the > regex is /^#.*(\d+)$/. Thus it didn't match, and $oneLine was never > re-initialized. > > # Port counters: Lid 8 port 1 (CapMask: 0x00) > PortSelect:......................1 > CounterSelect:...................0x0000 > SymbolErrorCounter:..............0 > LinkErrorRecoveryCounter:........0 > LinkDownedCounter:...............0 > PortRcvErrors:...................0 > PortRcvRemotePhysicalErrors:.....0 > PortRcvSwitchRelayErrors:........0 > PortXmitDiscards:................0 > PortXmitConstraintErrors:........0 > PortRcvConstraintErrors:.........0 > CounterSelect2:..................0x00 > LocalLinkIntegrityErrors:........0 > ExcessiveBufferOverrunErrors:....0 > VL15Dropped:.....................0 > PortXmitData:....................0 > PortRcvData:.....................0 > PortXmitPkts:....................0 > PortRcvPkts:.....................0 > > > On Fri, Mar 8, 2013 at 9:35 AM, Mark Seger <[email protected]> wrote: >> >> thanks joe. I looked at the code and am a little puzzled because >> $oneLine does get initialized every time collectl sees a record >> starting with /^#.*port \d+/ and even though what I have coded is >> clearly wrong, I've never seen the behavior you're describing. So if >> it is growing does that mean you're getting a different format record >> returned by your version of perfquery than everyone else who currently >> uses collectl to monitor their IB? I'd think that would be a bigger >> problem! >> >> In any event, clearly what I have is wrong and so it will be fixed in >> the next release which should be out in a week or 2. >> >> -mark >> >> On Mon, Mar 4, 2013 at 12:46 PM, Joe Digilio <[email protected]> >> wrote: >> > Attached is a patch to collectl 3.6.5. >> > >> > Collectl was creating huge files (>1GB/day and growing). It turns out >> > $oneLine was not getting cleared out before each check, so the "ib" line >> > was >> > growing longer and longer. >> > >> > -Joe >> > >> > >> > ------------------------------------------------------------------------------ >> > Everyone hates slow websites. So do we. >> > Make your web apps faster with AppDynamics >> > Download AppDynamics Lite for free today: >> > http://p.sf.net/sfu/appdyn_d2d_feb >> > _______________________________________________ >> > Collectl-interest mailing list >> > [email protected] >> > https://lists.sourceforge.net/lists/listinfo/collectl-interest >> > > > ------------------------------------------------------------------------------ Symantec Endpoint Protection 12 positioned as A LEADER in The Forrester Wave(TM): Endpoint Security, Q1 2013 and "remains a good choice" in the endpoint security space. For insight on selecting the right partner to tackle endpoint security challenges, access the full report. http://p.sf.net/sfu/symantec-dev2dev _______________________________________________ Collectl-interest mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/collectl-interest
