541012 - kmacinnis access + sudo: cm-hadoop0[7-9],cm-hadoop[12]*,cm-hadoop-adm*

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: Account Requests, task)

Description

[Ken MacInnis]

Per bz 536578 and 536580, the new production Hadoop machines are ready to go.  I am requesting access to these machines, as well as sudo, as with cm-hadoop0[1-6].

I do seem to have access to cm-hadoop-adm01 but not sudo.  I do not seem to have any access to the rest.

Comment 1

[Dave Miller [:justdave]]

The stuff I pushed last night didn't seem to be working this morning, and couldn't figure out why.  Turns out they needed to be rebooted after switching them to ldap auth.  I have a script going through rebooting them all now, 07, 08, 09, 12, and 18 should be usable now, the rest will be in the next half hour or so as they pick up the reboots.

Comment 2

[Dave Miller [:justdave]]

the bad nfs setup (which there's another bug somewhere for) is playing havoc with the reboot process because it takes like 20 minutes for the nfs mounts to time out, and it does that first before starting sshd or nrpe.

But as they come back up, everything should be working now, tested it with my personal account.

Comment 3

[Ken MacInnis]

My sudo access went away on cm-hadoop-adm03.  Could someone recheck this?

Also, my ssh keys seemed to disappear on the rest of the new machines, though I could recopy them over and things seem to be OK otherwise.  Perhaps related.

Comment 4

[Dave Miller [:justdave]]

Your ssh keys are in LDAP, it should be picking them up from there.  Local copies on the machine are subject to getting overridden by the configuration management stuff (which will put back the "official" copies any time it notices they've been changed), especially on the root account.

If you need a key on the root account for a specific purpose we'll need to add it via the configuration management.

sudo's been fixed on adm03, the puppetize script didn't finish on that box the other day when I ran it, and I had aborted that one when I discovered the machines were taking 20 minutes to reboot because of the NFS thing.  I had forgotted to go back and do it.  It's been updated now, should work the same as the others again.