7:30 - Doing the morning routine early today, drinking coffee and playing with our poor little confused cat Hiro.
- Hooper sent mail requesting the RHEL3 install media.� Replied with the RHN account to login and download the current media, and told him where the 3/6 media was in the office.
- Leaving home at 8 for the trek into the office by 9 for the meeting with Slick, Margie, et al. and the testing of Mage on the new switch.� Mage has 3 san luns assigned and will be a better test subject than Page was yesterday.� The primary function of the test is to see if the luns get out of order on reboot.
- I'm supposed to stop at AZE at 3:30 to see Sofya and the resident doctor.� I sincerely doubt I'll make it out of the office by 2:30, so...well.� whatever.� It's not as if it can be fixed at this point anyway.
- Pull a list together of all SMB mounts in the Portal and see how many are authenticating out of Redmond.� Correlate current and past mount issues to authentication master and see if there is a pattern.�
- Discuss the upgrade cycle from RHEL3 to 4 or 5 with Mr. T. today.� This will have to tie in with the openldap versions among other things.�
- Do calculations for the backup plan if LDAP rollout on r3/9 won't work.� Disk space requirements and time constraints to do the work.� ICD is pushing hard on Hoops.
12:55 - The mage test went okay this morning.� Successfully moved over to the new luns without issue, the data migration is running fine and we see no appreciable slowdown on the hosts.� DMP is working as advertised.
Chops buttonholed me regarding DMP on lavender.� While we see the multipath running (18 & 19) from lavender back to the dmx3, vxvm isn't reporting a multi path being up.� I talked this over with Mac and we decided to do an impromptu test of the situation by yanking the primary cable from lavender to see if it would failover.� It didn't.� We bounced the server with -r and it still didn't see it.� Mac and I are going to let Chops continue to work the issue since he discovered it, any resolution should be his kudos.� He'll find it, that kid is nothing if not tenacious.
Talked with Mr. T. regarding the upgrade (or the inability thereof) for our portal RHEL3 machines to 4.� We're going to wait for Becky to get back from medical leave to discuss.
2:00 - working on the SMB issues on the killian LM servers.� Started out on D.� Pointed them to Athens but it wouldn't join the domain complaining that there were no logon servers to service the request.� I specified the server and it worked:
net -d ads join -S -U
put in my password and the join completed.
Cycle winbind and smb but I can only browse to the websphere share.� I deleted the trans share since we're not using the transfer system here.�
The truly fucking bizarre part is that the [varlog] share wouldn't work as read-only.� It simply refused to share/browse and gave the error "incorrect function". Making it read/write worked. � When I changed the name to [varlogs] plural, it worked.� When I changed the name to [Varlog], singular with a capital V, it failed.� Leaving it as [varlog] failed.� Changing it to [varloga] worked.� I decided to set the share to read-only and set the name to [varlogs] and let it go.� I'm going to go work on the other killian lm server share and see if its just as fucko.
-- Need to ask Grandpa Simpson wtf is up with ns1.nonprod.� He built it as RHEL4/3 but on /dev/sdb instead of sda.� WTF did you do that for?� Does this have physical drives?� I don't think its booting from SAN.� WHY IN THE FUCK DID YOU SET NAMED LOGGING TO VERBOSE AND THEN ONLY GIVE IT A 1GB PARTITION FOR /VAR?� Jackhole.
Format the sda and put an LVM in there to expand /var.� please.� pretty please.� with sugar on top.
4:36 - last update for the day.� Moved over to killianS, the other box was killianD.� Copied the smb.conf from killianS to killianD, edited the krb5.conf file, took the extraneous entries out of hosts (STOP IT BECKY!), verified resolv.conf was pointing at fosters first, then letting fosters roll to external DNS as needed.� I set this file to share varlog as [varlog] and not [varlogs].
This time it worked as [varlog], but I ended up having to net join twice.� I also took out the password server entry in that file since authentication is set to ADS and not server.� You don't need the password server entry in that case.�
I also changed the NETBIOS name from KILLIAN-LM to KILLIAN-LMS.� This makes me wonder if the label [varlog] has to be unique if there is more than one server with the same NETBIOS name.� Seems like that would make sense.� I don't know how many servers we have with SMB mounts and are named killian.� (This is one of the drawbacks to LM, since LM allows for identical hostnames when that's normally strictly verboten)
-- Next week, edit killianS NETBIOS name and see if it'll let me share [varlog] without the 's'.
-- Fixed the login lag on Juno.� Shut down winbind and smb, there's no reason for them to be running there, I don't know wtf Grandpa Simpson was doing in the first place.
-- Don't forget to tell Chops about the NS1.nonprod plan next week.� He wants to know how I fixed Juno too, but maybe I'll let him stew till Monday.� lol.
-- Stopped at AZL to see the doctor.� She gave me a kit to work with.� Thrilling.� Have monday appt with Sofya.� I don't know why I'm doing this again.� 7PM, and only for 30 minutes.
- Scored 4 tix to the Mercury game on Tuesday night.� Auriel is going with me, I don't know who else to ask.� Perhaps my neighbor Donna would want to come along with Heather.
Add comment
