Hi, As a result of issues at an exchange point over the last few days, a number of us (ISPs) have noticed an issue with BGP sessions sitting in the "Idle" state, because the other end is shut down.
Basically, it appears that on Sup720s at least, once you reach a critical number of sessions in Idle (More than 5, less than 20) the CPU usage increases to 30%, all down to the BGP Router process. 30ish sessions down and it's up to 50% - we've had ours up to 70% as a result of this, although I don't know how many sessions were down at that point. This behaviour has been confirmed on 12.2(33)SRC4 and 12.2(33)SRD2, with other possible reports on SXF, SRC3, SRC5 and also on CRS-1s. Has anyone seen this before and know if it's a known issue with a BugID associated? A workaround is to apply "neigh x.x.x.x transport connection passive" but this clearly isn't optimal. _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
