On Mon, Oct 10, 2005 at 01:06:22PM +0200, Matthias Saou wrote: > Hi, > > All the RHEL4 servers I manage got updated this morning, with lots of > packages in the rpm transaction since they were running U1 and got all the > updates from U2. > > AFAICT, all the i386 servers updated fine, whereas all the x86_64 ones got > stuck with useradd taking up 100% CPU and impossible to kill even with > SIGTERM. From what I saw, it doesn't seem to be a package in particular, > since most were stuck on the nscd scriplet, but at least one was stuck on > the openssh scriplet. > > Even impossible to reboot the servers... they get stuck trying to kill the > process apparently, and a hard reboot was required. I tried to use "strace > -o useradd.log -p <pid>" on one of the running useradd processes, but it > gave an empty file and strace refused to stop with Ctrl+C, I had to kill > it. > > I still have one useradd command running on an x86_64 server, if anyone > has ideas on how to debug the problem. It's running at 100% CPU, not even > defunct... Can you identify whose package's script is doing that and what the full useradd command line looks like? Also lsof and lslk may help find out what the useradd process is choking on. -- Axel.Thimm at ATrpms.net
Attachment:
pgpTzGFwjizu0.pgp
Description: PGP signature