[mrtg] Massive peaks on counter resets

[Scott_Rochford at DELL.com]

Hi MRTG fans,

We use MRTG to monitor a variety of system performance counters
including Veritas Volume Manager stats from vxstat.  These are reset to
0 every night.

Previously using mrtg-2.9.25 this was working fine.  Since upgrading
recently to mrtg-2.13.0rc4 we have been getting massive peaks on the
charts every time the counter is reset.  What has changed between these
two versions that makes this happen?

We do NOT use rrdtool.

I have seen discussions of DERIVE versus COUNTER in the rrdtool list,
however they seemed to apply to rrdtool only.  I have also seen mention
of using 64-bit counters, however I can't see a way to turn that on for
values that are not collected using SNMP.

Here is a sample of the log file with such a peak:

1140100286 4614739 6217096
1140100286 1397 1658 1397 1658
1140099377 1105 1580 1105 1580
1140099300 1105 1580 1105 1580
1140099000 1105 1580 1105 1580
1140098700 1139 1577 1241 1580
1140098400 1241 1572 1241 1572
1140098100 1241 1572 1241 1572
1140097800 1246705 1263538 4614073 4675521
1140097500 4614073 4675521 4614073 4675521
1140097200 4614073 4675521 4614073 4675521
1140096900 3430080 3475828 4614073 4675521
1140096600 1115 1396 1115 1396
1140096300 1115 1396 1115 1396
1140096000 1246 1372 1628 1396
1140095700 1628 1305 1628 1305

Here are the settings in the MRTG config file for that parameter:

MaxBytes[_]: 999999999999
UnScaled[_]:
kMG[_]: k,M,G,T,P
kilo[_]: 1024
Options[_]: growright noinfo nopercent
YLegend[_]: Bytes / Sec
ShortLegend[_]: B/s
Legend1[_]: Total reads
Legend2[_]: Total writes
Legend3[_]: Peak reads
Legend4[_]: Peak writes
LegendI[_]:  reads:
LegendO[_]:  writes:
Colours[_]: SKYBLUE#0080FF, BLUE#000080, SKYBLUE#0080FF, BLUE#000080

Target[hostname_vxtotio_diskgroupname]: `/usr/bin/ssh hostname
/usr/local/bin/mrtgagent vxtot diskgroupname `
Title[hostname_vxtotio_diskgroupname]: Total I/O on Veritas Volumes in
diskgroupname
PageTop[hostname_vxtotio_diskgroupname]: <H1>Total I/O on Veritas
Volumes in diskgroupname</H1>

Any suggestions would be greatly appreciated!

--
Scott Rochford

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://lists.ee.ethz.ch/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi