Opened 11 years ago
Closed 11 years ago
#68 closed (fixed)
Clock wrong on some BBN IG nodes
Reported by: | jbs@bbn.com | Owned by: | somebody |
---|---|---|---|
Priority: | major | Milestone: | |
Component: | AM | Version: | SPIRAL5 |
Keywords: | Cc: | ||
Dependencies: |
Description
The clock is wrong on some BBN IG nodes:
[13:36:33] jbs@anubis:/home/jbs +$ shmux -c "date" {control,boss,ops,foam,flowvisor}.instageni.gpolab.bbn.com bbn-ig-jbs15 bbn-ig-jbs15: Wed Dec 19 07:35:07 EST 2012 foam.instageni.gpolab.bbn.com: Wed Dec 19 13:36:22 EST 2012 boss.instageni.gpolab.bbn.com: Wed Dec 19 13:36:34 EST 2012 ops.instageni.gpolab.bbn.com: Wed Dec 19 13:36:34 EST 2012 flowvisor.instageni.gpolab.bbn.com: Wed Dec 19 13:36:22 EST 2012 control.instageni.gpolab.bbn.com: Wed Dec 19 13:36:35 EST 2012 6 targets processed in 1 second. Summary: 6 successes
It's a little wrong on the FOAM and FV VMs, and more significantly wrong on my experiment VM.
Change History (8)
comment:1 Changed 11 years ago by
comment:2 Changed 11 years ago by
Leigh says:
From: Leigh Stoller <lbstoller@gmail.com> To: instageni-design@geni.net Subject: Re: [instageni-design] [GENI: instageni] #68: Clock wrong on some BBN IG nodes Date: Wed, 19 Dec 2012 13:07:07 -0800 > It's a little wrong on the FOAM and FV VMs, and more significantly wrong > on my experiment VM. So this is fixed on boss/ops and the shared hosts. I needed to beef up the ntp.conf file on ops, restart ntpd, wait for convergence (whatever that means, ntp is totally opaque to me), and then change all the times. I also had to change the timezone file on the control node; it was on mountain time. foam and flow visor will get to the right time eventually ... Lbs
Ok, looks better now, except on the FOAM and FV VMs. When you say that they'll get the right time "eventually", do you mean more like "when NTP automatically does its thing and synchs the clocks", or more like "when Nick sets up NTP on them", or ... ?
comment:3 Changed 11 years ago by
Leigh says:
From: Leigh Stoller <lbstoller@gmail.com> To: instageni-design@geni.net Subject: Re: [instageni-design] [GENI: instageni] #68: Clock wrong on some BBN IG nodes Date: Wed, 19 Dec 2012 13:13:27 -0800 > Ok, looks better now, except on the FOAM and FV VMs. When you say that > they'll get the right time "eventually", do you mean more like "when NTP > automatically does its thing and synchs the clocks", or more like "when > Nick sets up NTP on them", or ... ? I assume Nick is running ntp. Someone needs to log in and do the ntpdate thing, or just wait a long time as ntpd slowly changes the time (which is why you run ntpdate at boot, right?).
Well, beyond that, ntpd won't actually slowly change the time at all if it's off by too much to begin with -- it assumes that something has gone badly wrong, and that it had better just quietly do nothing.
In any case, it does not at a glance look like ntpd is running on the FOAM and FV VMs. Nick, bug or feature?
comment:4 Changed 11 years ago by
Resolution: | → fixed |
---|---|
Status: | new → closed |
ntpd is apparently not part of the default ubuntu server build for reasons passing understanding. It's in our build now (and enabled on foam and flowvisor).
comment:5 Changed 11 years ago by
Confirmed, things look good now:
[16:36:30] jbs@anubis:/home/jbs +$ shmux -c "date" {control,boss,ops,foam,flowvisor}.instageni.gpolab.bbn.com bbn-ig-jbs15 bbn-ig-jbs15: Wed Dec 19 16:37:28 EST 2012 foam.instageni.gpolab.bbn.com: Wed Dec 19 16:37:29 EST 2012 flowvisor.instageni.gpolab.bbn.com: Wed Dec 19 16:37:29 EST 2012 ops.instageni.gpolab.bbn.com: Wed Dec 19 16:37:29 EST 2012 boss.instageni.gpolab.bbn.com: Wed Dec 19 16:37:29 EST 2012 control.instageni.gpolab.bbn.com: Wed Dec 19 16:37:29 EST 2012 6 targets processed in 1 second. Summary: 6 successes
All set here.
comment:6 Changed 11 years ago by
Resolution: | fixed |
---|---|
Status: | closed → reopened |
comment:7 Changed 11 years ago by
Hmm, this seems to have drifted again on the shared nodes:
+$ shmux -c "date" {control,boss,ops,foam,flowvisor}.instageni.gpolab.bbn.com bbn-ig-jbs15 bbn-ig-jbs15: Thu Dec 27 19:09:04 EST 2012 foam.instageni.gpolab.bbn.com: Thu Dec 27 19:11:23 EST 2012 flowvisor.instageni.gpolab.bbn.com: Thu Dec 27 19:11:23 EST 2012 ops.instageni.gpolab.bbn.com: Thu Dec 27 19:11:23 EST 2012 boss.instageni.gpolab.bbn.com: Thu Dec 27 19:11:23 EST 2012 control.instageni.gpolab.bbn.com: Thu Dec 27 19:11:24 EST 2012 6 targets processed in 2 seconds. Summary: 6 successes
Did that change not stick?
comment:8 Changed 11 years ago by
Resolution: | → fixed |
---|---|
Status: | reopened → closed |
Leigh says:
From: Leigh Stoller <lbstoller@gmail.com> To: instageni-design@geni.net Subject: Re: [instageni-design] [GENI: instageni] #68: Clock wrong on some BBN IG nodes Date: Wed, 2 Jan 2013 11:04:49 -0800 > Did that change not stick? Sorta; ntpd core dumped on both pc1 and pc2. It is a known bug in the current (for Fedora15) version of ntpd, which I have worked around I think with a change to /etc/ntp.conf. I will need to rebuild the image in Utah and get the fix propagated to the Utah rack and Utah PG. The BBNs nodes are fixed.
A day later, things still look good. Thanks!
bbn-ig-jbs15 is: