cpsrvd crashing, any fixes yet?

[parasane]

The Xeon, goodmove. The AMD (XP 2400+) handled it like a champ, but the Xeons (dual, by the way) couldn't hack it.

[cpanelben]

We run Red Hat 9 with 2.6 kernels on most our servers. Seems to be a winning combination. I agree, you should be very cautious about changing a RHEL system to a 2.6 kernel; I guess I spoke out of haste.

On a good note though, I stumbled across another work around while investigating Bacula (Bacula User Guide). It seems they've been grappling with a similar issue, and they found that disabling the RH threading library proved to be an effective solution. I spoke briefly with one of Bacula's developers, and he said that he also had success with setting LD_ASSUME_KERNEL to an earlier version, but that just renaming the /lib/tls directory had the same affect and provided a more focused solution.

I also noticed that Red Hat released an upgrade to their glibc rpms for RHEL. How many of you have updated to the newer version?

And I find it odd that this issue didn't exist on the RHEL3 boxes until cpsrvd came along (back when the binaries were separate things were fine), and with all of the other services/processes running, only cpsrvd seems to have the issue. What specifically is cpsrvd exploiting that is triggering the so-called bug?

This issue became prevelant because of cpsrvd's greater reliance upon the threading library.