View previous topic :: View next topic |
Author |
Message |
Papillon x-man
Joined: 28 Dec 2004 Posts: 1060 Location: Germany
|
Posted: Tue Jun 14, 2005 12:22 Post subject: |
|
|
Hrmmm, why do you change the connection type ? I thought you said ODBC 0.9.2.4 would run fine ? I am mainly interested to know if changing just the .EXE makes a difference on your system. _________________ Papillon |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Tue Jun 14, 2005 16:04 Post subject: |
|
|
It did crash under ODBC connection with the newer NWNX.EXE, though it took longer.
I shifted to Direct Connect to see if maybe I just didn't leave it running long enough in the initial check under it on Sunday. |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Sat Jun 18, 2005 16:19 Post subject: |
|
|
Update Papillon...
It looks like I didn't wait long enough the first time I tried NWNX2.exe current under ODBC .9.2.4. Over the last week it has crashed 3 times under that setup, but takes 22-30 hours of running to do it. No interesting entries in the logs.
When it crashes, it does NOT cycle the log files; the previous run's logs are still primary. NWNX also stops writing to the log files, entries ARE made into those logs by NWServer. I cannot tell if it reads/writes to the database as it comes up with the "Port In Use" error attempting to restart. Since it does not appear to shift server port off of 5121 I almost have to assume that its the MySQL port that is locking.
I have NWNX set to check every 30 seconds with a 30 second delay on re-start.
ThanX! |
|
Back to top |
|
|
Papillon x-man
Joined: 28 Dec 2004 Posts: 1060 Location: Germany
|
Posted: Mon Jun 20, 2005 13:23 Post subject: |
|
|
So do I understand this correct:
* NWNX.EXE does not crash, but NWSERVER does
* NWNX Logfiles are not rotated when the server crashes
* You get a "port in use" error even with a restartdelay of, say, 60 seconds
* It only crashes when using the current NWNX.EXE, older versions of the exe work fine
* When using the older exe version, ODBC 0.9.2.4 still works fine with either connection type.
Please correct if I misunderstood anything. _________________ Papillon |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Mon Jun 20, 2005 16:28 Post subject: |
|
|
I have not set the restart delay for more than 30 seconds as previously the system never required more than 5 seconds to recover from the very rare crash (normally caused by my scripting....).
Other than that all points are correct. As for the log files, they do NOT rotate, but the new directories are added correctly.
I don't think you should put a lot of work into this as it seems to be isolated to my stuff I will not be able to work on this (or much else) for a couple of weeks due to family concerns but I am planning a rebuild of the server after that to upgrade hardware. Maybe that will help as the server has been running for over 2.5 years, though it has been well maintained. |
|
Back to top |
|
|
Papillon x-man
Joined: 28 Dec 2004 Posts: 1060 Location: Germany
|
Posted: Mon Jun 20, 2005 20:09 Post subject: |
|
|
Ok, thanks for your input. What I will do is to go over the source code and check the changes that have been made between the old and the new nwnx main program. Maybe I will be able to see something, but probably it is just some strange occurence pertaining just to your server.
As much as I would like to solve your problem, if you do not have enough time to conduct tests I am caught flat-fooded with a critical strike _________________ Papillon |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Mon Jun 20, 2005 20:50 Post subject: |
|
|
ThanX Papillon, you've done more than enough already!
I'll report back as for how things work after the reload, I'm thinking that will probably correct it as no one else seems to be having this issue |
|
Back to top |
|
|
Lokey
Joined: 02 Jan 2005 Posts: 158
|
Posted: Wed Jun 22, 2005 1:42 Post subject: |
|
|
Port in use and fails to rotate the logs--sounds familiar. Go over your permission settings with a fine tooth comb...read about our team's trials with Server2k3 here. _________________ Neversummer PW NWNx powered mayhem |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Wed Jun 22, 2005 2:01 Post subject: |
|
|
But only after a crash?
Combined with NWNX 2.5.3 being stable and running for literally weeks under a light load its not permissions. When NWServer does crash under 2.5.3 it always starts back up correctly... |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Tue Jun 28, 2005 1:38 Post subject: |
|
|
Update:
Saturday I converted to the SQLite database. 43 hours uptime with NWNX2 2.6.1 and ODBC 9.2.4 before I had to shut it down to fix a faction issue. I will report back if anything develops but it does look like something with MySQL. |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Thu Jun 30, 2005 6:19 Post subject: |
|
|
False hope. Its been crashing every 4-6 hours under load
I installed a fresh copy of the dedicated server, downloaded fresh copies of all the DLLs and no change in behavior.
We will see how it works after a reload. |
|
Back to top |
|
|
Divinity
Joined: 11 Jul 2005 Posts: 4
|
Posted: Mon Jul 11, 2005 1:08 Post subject: |
|
|
Papillon wrote: | The root of this problem might be hard to find. If I were you, I would not wait for a new version of NWNX that magically solves all troubles - since I do not know what is causing the problems on your system, I can do nothing to fix it.
The oddest thing is the 30% CPU usage that you see. Does that happen with nwnx and the odbc plugin alone (i.e. with no other plugins active) ? Does it happen with the aps_demo module ? Did you check the profiler for any apparent script problems (that maybe only occur with the current nwnx) ? Also you could enable the highest profiler logging option to see what scripts have been running last, before a crash occurs.
If you are running ODBC 0.8.8 it is already very close to the the current version, minus the SCORCO hooks, btw, that makes it even stranger. Please try to enable the scorco hooks and see if that changes anything. You could also temporarily use SQLite instead of MySQL.
These are just some ideas from the top of my head. I am sure the problem can be found, but it requires some assistance from your side. |
Indeed, it sounds like your "hourglass" is being suffocated. 33cpu usage when empty is quite rediculously alarming. Sounds nothing like a nwnx or nwnserver bug. if it would be my guess there are other programs running deep in your registries walls or what not. Good luck |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Mon Jul 11, 2005 1:36 Post subject: |
|
|
Divinity wrote: |
Indeed, it sounds like your "hourglass" is being suffocated. 33cpu usage when empty is quite rediculously alarming. Sounds nothing like a nwnx or nwnserver bug. if it would be my guess there are other programs running deep in your registries walls or what not. Good luck |
But it was only happenign with NWNX current. Previous versions did not show this behavior. And I am well-versed in cleaning/maintaining machines that have aquired interesting bugs and such as I am a sys admin for a school district with 16 buildings, 38 servers and 2900 desktop machines
Since this was posted I completely reloaded the server, downloaded fresh copies of the dedicated server and new copies of all the DLLs. Server still exhibits this behavior
I have a test server up and running under evaluation that at this time (three days time) appears more stable though it hasn't been long enough or under heavy enough of a load to be certain. Its completely different hardware (even different brand of hard drives) and different OS.
I am planning on throwing the previous machine on our hardware diagnostic equipment at work and seeing if it is actually a hardware issue. If the hardware tests good I'll start elemination testing on the drivers
Hopefully I will be able to determine exactly what the issue is so I can give Papillon more information if it is some kind of odd conflict and not just wear and tear on the hardware! |
|
Back to top |
|
|
Acrodania
Joined: 02 Jan 2005 Posts: 208
|
Posted: Thu Jul 14, 2005 16:28 Post subject: |
|
|
Update:
Ran hardware and software diagnostics on server. No errors noted, including memory. Changed Video and Network to cards that use different chipsets. Machine has no sound so couldn't alter that Current NIC is a 3C905C-TX.
Reloaded Win2K from scratch, patched to current. Loaded different VNC server, different FTP server. Moved entire directory of NWNServer from test server (which had been running as long as 90 hours between module updates).
Machine runs between 4.5 and 5.5 hours between NWNServer crashes, same as before. NWNX is set to check every 120 seconds, with a restart after 360 seconds. When server comes back up it still has the "Port open" error, its IP address shows as "XXX.XXX.XXX.XXX:-1". Server does NOT have the higher CPU useage it showed earlier under NWNX 2.61/ODBC 0.9.2
This continued crash issue with the same symptoms leads me to assume that its either a conflict with the motherboard drivers or something in the way I load Win2K.
Has anyone had problems with the Intel 845 main boards? The stable test server was running on an P3 MSI board (under Win98 ) |
|
Back to top |
|
|
|