My geek thread
04-05-11, 12:07 PM
#601
My idea of fun
Join Date: Jun 2006
Location: Gainesville, FL
Posts: 9,920
Bikes: '06 Litespeed Tuscany, '02 Kona Lavadome, '07 Giant TCR Advanced, '07 Karate Monkey
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 41 Post(s)
Liked 59 Times
in
36 Posts
Six years ago I switched my back end printing system to cups when I switched my primary back-end server from solaris 8 to freebsd. During that switch, an issue crept up where solaris clients printing to it (via the cups-lpd inetd subsystem) always printed a banner page.
Since we'd shrunk down to just two solaris workstations in the department, and only one of them does any printing (it's the client that runs "The Publisher" for the former chair), it was a back-burner issue for me. Sure, I worked on finding a solution for awhile, but eventually gave up and never looked back. It really wasn't a big deal to me that one user printed banner pages to his own personal printer, and he usually printed from a linux box anyway, so the banner pages only cropped up when he printed historical documents from "The Publisher" (I can go on another rant about The Publisher, but I'll pass)..
So I've been migrating all of my servers from freebsd to ubuntu lts. I've already migrated my other servers (web, bastion host, mail, database, tomcat/jakarta servers). I only have this one final server, and it's the one I'm sweating bullets on. It's our NIS master, NFS server, samba, cups, lan manager, and a few other things. I'm shuttering it on May 3rd, but I want to make sure all of the pieces are in place before that migration happens.
In the process of working on the migration for that final box, I tested the cups-lpd printing today from the sun client to the new server and re-discovered the banner page bug. I had completely forgotten about it, out of sight, out of mind I guess.
Well, I decided to google the issue. Back in 2006 (the last time I searched for a solution), I couldn't find an answer. Now, with five years having passed, I found an answer from 2007. And it works! The banner page is gone. The former chair will probably be surprised, but happy to see his banner pages are gone.
So, a five year old bug has been quashed as part of this migration. Fear my foo.
This also reminds me that I am being pretty thorough on this migration, so maybe, just maybe, I won't miss anything...
Since we'd shrunk down to just two solaris workstations in the department, and only one of them does any printing (it's the client that runs "The Publisher" for the former chair), it was a back-burner issue for me. Sure, I worked on finding a solution for awhile, but eventually gave up and never looked back. It really wasn't a big deal to me that one user printed banner pages to his own personal printer, and he usually printed from a linux box anyway, so the banner pages only cropped up when he printed historical documents from "The Publisher" (I can go on another rant about The Publisher, but I'll pass)..
So I've been migrating all of my servers from freebsd to ubuntu lts. I've already migrated my other servers (web, bastion host, mail, database, tomcat/jakarta servers). I only have this one final server, and it's the one I'm sweating bullets on. It's our NIS master, NFS server, samba, cups, lan manager, and a few other things. I'm shuttering it on May 3rd, but I want to make sure all of the pieces are in place before that migration happens.
In the process of working on the migration for that final box, I tested the cups-lpd printing today from the sun client to the new server and re-discovered the banner page bug. I had completely forgotten about it, out of sight, out of mind I guess.
Well, I decided to google the issue. Back in 2006 (the last time I searched for a solution), I couldn't find an answer. Now, with five years having passed, I found an answer from 2007. And it works! The banner page is gone. The former chair will probably be surprised, but happy to see his banner pages are gone.
So, a five year old bug has been quashed as part of this migration. Fear my foo.
This also reminds me that I am being pretty thorough on this migration, so maybe, just maybe, I won't miss anything...
04-05-11, 12:27 PM
#602
Making a kilometer blurry
Nope, it was the date. March 31st. I walked into the lab with the whole team and three levels of management looking at me to debug this. Right as I walked in the door, I realized it had to be the date.
Turns out that someone had written some code to add three months to the current date. Since there's no June 31st, the server's calendar module freaked. I fixed it by converting the date to milliseconds and adding 90 days worth of milliseconds to that, then converting it back to a calendar date.
good stuff
Turns out that someone had written some code to add three months to the current date. Since there's no June 31st, the server's calendar module freaked. I fixed it by converting the date to milliseconds and adding 90 days worth of milliseconds to that, then converting it back to a calendar date.
good stuff
04-05-11, 12:28 PM
#603
Senior Member
Join Date: Jun 2010
Location: Western MA
Posts: 15,669
Bikes: Yes
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Likes: 0
Liked 0 Times
in
0 Posts
Best of luck, kensuf.
Of the things you listed, you can reduce the risk with a CNAME. The thing I would be most edgy about would be the NIS master and the ability to create, update, and push maps. You can't test that ahead of time in the same domain.
Of the things you listed, you can reduce the risk with a CNAME. The thing I would be most edgy about would be the NIS master and the ability to create, update, and push maps. You can't test that ahead of time in the same domain.
04-05-11, 12:59 PM
#604
My idea of fun
Join Date: Jun 2006
Location: Gainesville, FL
Posts: 9,920
Bikes: '06 Litespeed Tuscany, '02 Kona Lavadome, '07 Giant TCR Advanced, '07 Karate Monkey
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 41 Post(s)
Liked 59 Times
in
36 Posts
Oh yeah, I'm not worried about the NIS portion at all (except looking forward to ditching the master.passwd bsd anachronism). I've already tested that stuff out, I'm now down to the nitty-gritty (like printing from the sun), and maintenance scripts.
04-06-11, 12:03 PM
#605
My idea of fun
Join Date: Jun 2006
Location: Gainesville, FL
Posts: 9,920
Bikes: '06 Litespeed Tuscany, '02 Kona Lavadome, '07 Giant TCR Advanced, '07 Karate Monkey
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 41 Post(s)
Liked 59 Times
in
36 Posts
not all filesystems are created equal. ufs > ext(2,3,4). Just sayin'.
04-06-11, 01:53 PM
#606
Genetics have failed me
CRC errors in the SAN fabric on those stupid IBM HS22 BladeCenter has cost me about 4 weeks of solid trainer/interval time. Working till 10pm up to past midnight has drained my motivation to even make me look at my bike. 
Maybe this should go into the training failure thread...
Maybe this should go into the training failure thread...
__________________
Gelato aficionado.
Gelato aficionado.
04-06-11, 06:22 PM
#607
Senior Member
Join Date: Jun 2010
Location: Western MA
Posts: 15,669
Bikes: Yes
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Likes: 0
Liked 0 Times
in
0 Posts
04-07-11, 11:53 AM
#611
Genetics have failed me
Unknown at this point. HBA's were swapped out from QLogic to Emulex, since IBM mentioned that the CRC errors from the QLogic are a known problem on the BladeCenter (yet, they still sell that **** as an approved config)
Also, there are no fiber cables between the blades and the Brocade switch inside the chassis. And the switch has been swapped out with the same results. And only some blades complain about CRC errors, not all of them. And only in a very random fashion. It does work reasonable well at 4Gb speeds tho, 8Gb speeds on the other hand results in lotsa CRC errors.
Also trying to hook up a Hitachi VSP directly to those BladeCenter switches at 8Gb is also a no go, as it doesn't even connect. At 4Gb the connection works fine. Connecting the VSP to a DCX-4S 48port 8Gb blade results in a proper and very stable 8Gb connection to the VSP.
We've filed a former complaint with IBM and haven't heard any updates that they found a fix so far.
If you plan on going 8Gb and IBM BladeCenter HS22's, think twice before you purchase.
Also, there are no fiber cables between the blades and the Brocade switch inside the chassis. And the switch has been swapped out with the same results. And only some blades complain about CRC errors, not all of them. And only in a very random fashion. It does work reasonable well at 4Gb speeds tho, 8Gb speeds on the other hand results in lotsa CRC errors.
Also trying to hook up a Hitachi VSP directly to those BladeCenter switches at 8Gb is also a no go, as it doesn't even connect. At 4Gb the connection works fine. Connecting the VSP to a DCX-4S 48port 8Gb blade results in a proper and very stable 8Gb connection to the VSP.
We've filed a former complaint with IBM and haven't heard any updates that they found a fix so far.
If you plan on going 8Gb and IBM BladeCenter HS22's, think twice before you purchase.
__________________
Gelato aficionado.
Gelato aficionado.
04-07-11, 01:10 PM
#612
Senior Member
Join Date: Jun 2010
Location: Western MA
Posts: 15,669
Bikes: Yes
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Likes: 0
Liked 0 Times
in
0 Posts
I managed a project a few years back where we built a 160 node grid using BladeCenters. We had a lot of trouble connecting the D-Link switch modules cleanly to the Cisco switches. It took several revs of firmware to fix, and we had to use workarounds in production until they were available. So I am not surprised by your story.
05-24-11, 08:44 PM
#615
negligent.
Join Date: Aug 2005
Location: White Plains, NY
Posts: 837
Bikes: a few
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Likes: 0
Liked 0 Times
in
0 Posts
hey geeks, wake up...what's your "favorite" (aka most-hated) compiler, linker, or runtime error?
a co-worker asked me to look at a bug of his today, I said sure. THEN I check the core dump.
"pure virtual method called"
arrrrrgh. I should have asked what the problem was before accepting!
a co-worker asked me to look at a bug of his today, I said sure. THEN I check the core dump.
"pure virtual method called"
arrrrrgh. I should have asked what the problem was before accepting!
05-25-11, 05:59 AM
#616
Senior Member
Join Date: Jun 2010
Location: Western MA
Posts: 15,669
Bikes: Yes
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Likes: 0
Liked 0 Times
in
0 Posts
None of the above.
STOP 0x0000abcde "Windows has experienced an unrecoverable error"
kernel panic: <insert martian speak here>
STOP 0x0000abcde "Windows has experienced an unrecoverable error"
kernel panic: <insert martian speak here>
05-25-11, 06:02 AM
#617
Elite Fred
Thread Starter
Join Date: Aug 2005
Location: Edge City
Posts: 10,945
Bikes: 2009 Spooky (cracked frame), 2006 Curtlo, 2002 Lemond (current race bike) Zurich, 1987 Serotta Colorado, 1986 Cannondale for commuting, a 1984 Cannondale on loan to my son
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 60 Post(s)
Liked 42 Times
in
19 Posts
Gotta put some Homestarrunner in here:
"FLAGRANT SYSTEM ERROR
Computer over.
Virus = Very Yes."
"FLAGRANT SYSTEM ERROR
Computer over.
Virus = Very Yes."
05-25-11, 07:54 AM
#618
Senior Member
Join Date: May 2007
Location: Santa Cruz Mountains
Posts: 6,169
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Likes: 0
Liked 1 Time
in
1 Post
My favorite unix error code is ENOTTY- "device is not a tty". It's so quaint.
But "Abort, Retry, Fail" has to be the dumbest error dialog ever. I'd rather get a kernel panic core dump- I can read Martian.
But "Abort, Retry, Fail" has to be the dumbest error dialog ever. I'd rather get a kernel panic core dump- I can read Martian.
05-25-11, 09:35 AM
#620
Making a kilometer blurry
I hate the ambiguous ones. From Visual Studio, for example:
"Some of the properties associated with the solution could not be read"
Thanks. That's just great.
"Some of the properties associated with the solution could not be read"
Thanks. That's just great.
05-25-11, 10:01 AM
#621
My idea of fun
Join Date: Jun 2006
Location: Gainesville, FL
Posts: 9,920
Bikes: '06 Litespeed Tuscany, '02 Kona Lavadome, '07 Giant TCR Advanced, '07 Karate Monkey
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 41 Post(s)
Liked 59 Times
in
36 Posts
'guru meditation error' anyone?
05-25-11, 10:15 AM
#622
Making a kilometer blurry
Here's my favorite API call:
https://developer.android.com/referen...ang.Throwable)
https://developer.android.com/referen...ang.Throwable)
Code:
public static int wtf (String tag, Throwable tr) Since: API Level 8 What a Terrible Failure: Report an exception that should never happen. Similar to wtf(String, String), with an exception to log. Parameters tag Used to identify the source of a log message. tr An exception to log.
05-25-11, 10:55 AM
#623
Elite Fred
Thread Starter
Join Date: Aug 2005
Location: Edge City
Posts: 10,945
Bikes: 2009 Spooky (cracked frame), 2006 Curtlo, 2002 Lemond (current race bike) Zurich, 1987 Serotta Colorado, 1986 Cannondale for commuting, a 1984 Cannondale on loan to my son
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 60 Post(s)
Liked 42 Times
in
19 Posts
wtf == "What a Terrible Failure"
05-25-11, 10:59 AM
#624
**** that
Join Date: Dec 2006
Location: CALI
Posts: 15,402
Mentioned: 151 Post(s)
Tagged: 0 Thread(s)
Quoted: 1099 Post(s)
Liked 104 Times
in
30 Posts
hey geeks, wake up...what's your "favorite" (aka most-hated) compiler, linker, or runtime error?
a co-worker asked me to look at a bug of his today, I said sure. THEN I check the core dump.
"pure virtual method called"
arrrrrgh. I should have asked what the problem was before accepting!
a co-worker asked me to look at a bug of his today, I said sure. THEN I check the core dump.
"pure virtual method called"
arrrrrgh. I should have asked what the problem was before accepting!
Fun game. I think my fave is "ERR_OK" from the COM world. It's an error code for when everything is ok.
Wow, what is that one from?
Here's my favorite API call:
https://developer.android.com/referen...ang.Throwable)
https://developer.android.com/referen...ang.Throwable)
Code:
public static int wtf (String tag, Throwable tr) Since: API Level 8 What a Terrible Failure: Report an exception that should never happen. Similar to wtf(String, String), with an exception to log. Parameters tag Used to identify the source of a log message. tr An exception to log.
05-25-11, 11:34 AM
#625
My idea of fun
Join Date: Jun 2006
Location: Gainesville, FL
Posts: 9,920
Bikes: '06 Litespeed Tuscany, '02 Kona Lavadome, '07 Giant TCR Advanced, '07 Karate Monkey
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 41 Post(s)
Liked 59 Times
in
36 Posts
Here's a screen cap of the error:
Here's the background: https://en.wikipedia.org/wiki/Guru_Meditation
Here's the background: https://en.wikipedia.org/wiki/Guru_Meditation