Question still stands, but I think I found my leak. Remember how we just discovered that new table that had, among other things, idle time in it? Well, the leak seems to come from walking that table. Sigh. I went back through everything I touched on the chassis in the last day and a half, and realized it was probably my newly-rewritten 'arcwho' utility. (I had thought OSPF because I had made a change yesterday... but only on two chassis, and the leak showed up on *all* of them, including some I hadn't touched in weeks. So it's not OSPF.) Here's the leak: if I snmpwalk the uumActiveSessionIfName tree, .1.3.6.1.4.1.429.4.2.1.46.1.6 (interface names of current sessions), I leak a few KB of free memory per walk. If I walk .46.1.7 (ip addresses), .46.1.8 (start time), or .46.1.9 (idle time), I do NOT leak memory. It's a shame I didn't know about this table in the beta, 'cause I probably would have found the leak then instead of on a Friday night. Bummer. I can't find a "disable snmp server/enable snmp server" type of command to try to clear memory usage either. More bummer. If a fix is released, I won't be able to download it 'cause I don't have a contract. More more bummer. Mike Andrews (MA12) * mandrews@dcr.net * http://www.bit0.com/ VP, sysadmin, & network guy, Digital Crescent Inc, Frankfort KY Internet services for Frankfort, Lawrenceburg, Owenton, Shelbyville www.fark.com: If it's not news, it's Fark. (Or something like that.) On Fri, 9 Jun 2000, Mike Andrews wrote:
What's the best way to track down a really big memory leak?
"Big" = enough to crash a 64 meg ARC in 24 hours.
law-ts1> show mem util
SYSTEM MEMORY RESOURCES Total System Memory Resources: 49509 KB Code Size: 5833 KB Initialized Data Size: 1490 KB Uninitialized Data Size: 4356 KB Stack Size: 512 KB Free Memory Current Value: 13451 KB Free Memory 1 Hour Before: 14590 KB Free Memory 12 Hours Before: 20724 KB Free Memory 24 Hours Before: 19901 KB Total Buffer Cache: 40000 Free Buffer Cache: 26690
Almost a meg an hour. Ouch.
This just started happening about 28 hours ago, probably when I changed something in the config... hell if I remember what it was though. OSPF maybe. It's kinda like what happens to a NETserver when syslog is broken, but I don't think syslog is the culprit here...
Is there some way I can get a list of running processes (like "list processes" does) on the card and how big each one is? I've got a list of all commands in 5.0, including the "_" commands, but so far I haven't found a winner.
Mike Andrews (MA12) * mandrews@dcr.net * http://www.bit0.com/ VP, sysadmin, & network guy, Digital Crescent Inc, Frankfort KY Internet services for Frankfort, Lawrenceburg, Owenton, Shelbyville www.fark.com: If it's not news, it's Fark. (Or something like that.)
- To unsubscribe to usr-tc, send an email to "majordomo@xmission.com" with "unsubscribe usr-tc" in the body of the message. For information on digests or retrieving files and old messages send "help" to the same address. Do not use quotes in your message.
- To unsubscribe to usr-tc, send an email to "majordomo@xmission.com" with "unsubscribe usr-tc" in the body of the message. For information on digests or retrieving files and old messages send "help" to the same address. Do not use quotes in your message.