Hostmon Out of System Resources
Hostmon Out of System Resources
I'm running version 5.12 on Windows 2003 server, with 4-3.20GHz cpu's and 4Gig's of memory, with 1100 tests, mostly PING and UNC tests, running as a service. Each time an alert is generated I use an external Trapgen program to send a SNMP trap to a Unix trap server. After a few days I try to edit a test or profile or perform any task which opens a sub-window and receive an error message saying that the process can not be completed, out of system resources. Once I stop and restart the service it works fine for the next few days. This sounds like you still have the memory leak bug that was fixed in ver. 5.10.
I have a few URL tests and one service test. I don't use any ODBC logging.
System Resources show: GDI Objects=391, User Objects=211, Memory=19K, Handles=200, Threads=11.
No anti-virus software installed.
I can take a snapshot of the window the next time it happens, but how do I send you a .jpg or .doc file?
System Resources show: GDI Objects=391, User Objects=211, Memory=19K, Handles=200, Threads=11.
No anti-virus software installed.
I can take a snapshot of the window the next time it happens, but how do I send you a .jpg or .doc file?
Everything is working fine now. When it happens again, I will take snapshots and send them to you.
BTW - Your snmp trap alerts only let me send one varbind per trap. With the trapgen program that I use, I'm able to send multiple varbinds in one trap. Example: "c:\program files\trapgen\trapgen" -d 14.17.70.214 -c S3CCCT -s 108025 -o 1.3.6.1.4.1.3471 -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%TestName%" -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%Status%" -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%TaskComment%" -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%TestMethod%"
BTW - Your snmp trap alerts only let me send one varbind per trap. With the trapgen program that I use, I'm able to send multiple varbinds in one trap. Example: "c:\program files\trapgen\trapgen" -d 14.17.70.214 -c S3CCCT -s 108025 -o 1.3.6.1.4.1.3471 -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%TestName%" -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%Status%" -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%TaskComment%" -v 1.3.6.1.4.1.11.2.17.2.4 STRING "%TestMethod%"
Out of memory
We also see more and more 'out of memory' errors. This happens only when a lot of reports needs to be generated (we don't use external programs, so I can't confirm that this is the same / similar problem but it does look like it). It does hower not seem to be a lack of memory on the system part, since a few hundreds MB real memory are still available when we see this problem occur and the pagefile is far from being full.
Since the message we get are HostMonitor messages (no system messages indicating a lack of memory), it seems to be a HostMonitor problem. Also HostMonitor does not crash or stop, it just runs on generating more and more out of memory errors when a report needs to be created. It looks like the report engine is sometimes a bit confused
Since the message we get are HostMonitor messages (no system messages indicating a lack of memory), it seems to be a HostMonitor problem. Also HostMonitor does not crash or stop, it just runs on generating more and more out of memory errors when a report needs to be created. It looks like the report engine is sometimes a bit confused

I do run alot of reports. Whenever a tests fails, I write it to a report. Also I have reports setup that run every minute. I have trace route tests running and I create reports for them every 5 minutes. I have URL tests which check response time and site availability, which I also write the results to reports.
Alex: Just FYI, like Marcus, I've also had the "Out Of Memory" errors occur --- on 3 occassions -- typically after the HM machine has been running constantly for two weeks or more. This could indicate a very small/slow mem leak but because it's not been a huge issue, I haven't spent much time pursuing it. I've bigger fish to fry at the moment.
We don't have a lot of reports, but the ones we do have could be where this error is coming from. It appears Marcus has a better feel for this than I do.
We don't have a lot of reports, but the ones we do have could be where this error is coming from. It appears Marcus has a better feel for this than I do.
Today I tested HostMonitor on Windows 2000 SP4, generated a lot of reports but do not see any resource leakage.
Probably there is error in some system DLL... Will try on Windows Server 2003.
Timn, Marcus: what Windows do you use? Windows 2003 as well?
IMHO Windows 2000 is the best Windows, Windows 2003 has some bugs, Windows XP has more bugs (I do not count NT 4.0 - its too old and Windows 9x that is not real multitasking OS).
Regards
Alex
Probably there is error in some system DLL... Will try on Windows Server 2003.
Timn, Marcus: what Windows do you use? Windows 2003 as well?
IMHO Windows 2000 is the best Windows, Windows 2003 has some bugs, Windows XP has more bugs (I do not count NT 4.0 - its too old and Windows 9x that is not real multitasking OS).
Regards
Alex
We also use Win2K SP4 - same as Marcos. In our environment, HM is running on it's own dedicated box.
These events are very rare for us. When they do happen, the "Out of Memory" messages do seem to be coming from HM rather than from the OS
Next time it happens, I'll try to gather some additional info.
These events are very rare for us. When they do happen, the "Out of Memory" messages do seem to be coming from HM rather than from the OS
Next time it happens, I'll try to gather some additional info.
For us, it only seems to occur after HM has been running continuously for at least two weeks - but even then the problem is rare. For example, HM has currently been running for nearly 4 weeks with no problems.
We have roughly 3,000 tests with an estimated load of 19 tests/sec. We have only 4 different reports.
The problem is so infrequent for us that it is really 'below our radar'. We have other applications we wished ran as well as HM.
We have roughly 3,000 tests with an estimated load of 19 tests/sec. We have only 4 different reports.
The problem is so infrequent for us that it is really 'below our radar'. We have other applications we wished ran as well as HM.