Sunday, September 16, 2018

Cisco 3850 memory spike issue

Hey guys,

Is it just me or anyone else also think old Cisco switches were more stable? Since we've replaced our 3750s with 3850s we've been having issues with them. We've replaced two so far as they were faulty and now we're facing this issue with the brand new replacement one!

Last week, we had this weird issue that one of our 3850 switch stack went crazy with memory and became unresponsive. Luckily network didn't go down but we couldn't get to the console or via ssh. Eventually it resolved itself and memory went back to normal. Cisco TAC has no idea so far and just going through the logs. Have you seen an issue like this ?

IOS: Cisco IOS XE Software, Version 16.06.02

Model Number : WS-C3850-48P

There were heaps of logs like below:

4801: Alternate Pool: None Free: 0 Cause: No Alternate pool

4802: -Process= "IOSD ipc task", ipl= 0, pid= 74

4800: Pool: Processor Free: 1598872 Cause: Memory fragmentation

4799: Memory allocation of 3380 bytes failed from 0xAAB9E1E5A0, alignment 8

4798: -Traceback= 1#bfcda4e05e635ef59d927d2feb906f16 :AAB316B000+4A0EAF0 :AAB316B000+4A0DC5C :AAB316B000+4A0E00C :AAB316B000+6C9A4CC :AAB316B000+6C72CA8

4797: Could not expand chunk pool for messages. No memory available -Process= "Chunk Manager", ipl= 2, pid= 1

4796: -Traceback= 1#bfcda4e05e635ef59d927d2feb906f16 :AAB316B000+4A0EAF0 :AAB316B000+4A0DC5C :AAB316B000+4A0E00C :AAB316B000+3A66B70 :AAB316B000+3A08AA8 :AAB316B000+39E89A8 :AAB316B000+3A0A234 :AAB316B000+3A0B0C4 :AAB316B000+3A36DB8 :AAB316B000+3A6877C :AAB316B000+6E5F6A4

4795: Internal error, Invalid event State :MAC-REACHABLE, Event :MAC_VERIFIED -Process= "SISF Main Thread", ipl= 0, pid= 505

4793: Could not expand chunk pool for messages. No memory available -Process= "Chunk Manager", ipl= 2, pid= 1

4794: -Traceback= 1#bfcda4e05e635ef59d927d2feb906f16 :AAB316B000+4A0EAF0 :AAB316B000+4A0DC5C :AAB316B000+4A0E00C :AAB316B000+6C9A4CC :AAB316B000+6C72CA8



No comments:

Post a Comment