Queue Manager - AMX Webserver issue

a_riot42a_riot42 AMX WizardPosts: 1,619
I am throwing this out there to see if anyone has else has run across this. Using a 3100 with 3.50.430 with a gateway connected also with the latest firmware, whenever I access the 3100 webserver, I start to see the Queue manager start to print error messages in diagnostics incrementing the size of the queue by 25 until it finally overflows. There is no way to get it to stop other than a reboot and the master becomes completely unresponsive. After a reboot all is well until I access the masters webserver. Anyone seen anything like this?
Paul

Comments

  • ericmedleyericmedley Senior Member - 3709 Posts Posts: 4,152
    a_riot42 wrote: »
    I am throwing this out there to see if anyone has else has run across this. Using a 3100 with 3.50.430 with a gateway connected also with the latest firmware, whenever I access the 3100 webserver, I start to see the Queue manager start to print error messages in diagnostics incrementing the size of the queue by 25 until it finally overflows. There is no way to get it to stop other than a reboot and the master becomes completely unresponsive. After a reboot all is well until I access the masters webserver. Anyone seen anything like this?
    Paul

    Just out of curiosity, what browser are you using to get there? I hope this doesn't disolve into a 'My browser can beat up your browser's dad" discussion. But I wonder if it may have something to do with the issue. I know the door bells and whatno don't like IE.
  • a_riot42a_riot42 AMX Wizard Posts: 1,619
    ericmedley wrote: »
    Just out of curiosity, what browser are you using to get there? I hope this doesn't disolve into a 'My browser can beat up your browser's dad" discussion. But I wonder if it may have something to do with the issue. I know the door bells and whatno don't like IE.

    Firefox 3.5.8. I typically don't use IE except for OWA so I don't know if it will do it with IE as well.
    Paul
  • yuriyuri Junior Member Posts: 861
    offtopic:

    Only for OWA?
    Then check this, it works like a charm:
    https://addons.mozilla.org/nl/firefox/addon/1419
  • a_riot42a_riot42 AMX Wizard Posts: 1,619
    yuri wrote: »
    offtopic:

    Only for OWA?
    Then check this, it works like a charm:
    https://addons.mozilla.org/nl/firefox/addon/1419

    Does it allow for different proxies depending on whether you are using a IE or FF tab? Looks pretty cool though, I might give it a try.
    Paul
  • tracktoystracktoys Junior Member Posts: 46
    Line 196 (11:05:27):: Accepted Telnet connection:socket=29 addr=10.77.0.110 port=54507
    Line 197 (11:05:32):: CIpEvent::AddInternalEvent - Max Queue Count = 125
    Line 198 (11:05:33):: CIpEvent::AddInternalEvent - Max Queue Count = 150
    Line 199 (11:05:34):: CIpEvent::AddInternalEvent - Max Queue Count = 175
    Line 200 (11:05:35):: CIpEvent::AddInternalEvent - Max Queue Count = 200
    Line 201 (11:05:36):: CIpEvent::AddInternalEvent - Max Queue Count = 225
    Line 202 (11:05:37):: CIpEvent::AddInternalEvent - Max Queue Count = 250
    Line 203 (11:05:38):: CIpEvent::AddInternalEvent - Max Queue Count = 275
    Line 204 (11:05:39):: CIpEvent::AddInternalEvent - Max Queue Count = 300
    Line 205 (11:05:39):: CIpEvent::AddInternalEvent - Max Queue Count = 325
    Line 206 (11:05:40):: CIpEvent::AddInternalEvent - Max Queue Count = 350
    Line 207 (11:05:41):: CIpEvent::AddInternalEvent - Max Queue Count = 375


    I just noticed this very same issue when accessing the master through telnet within NS. This happened with the current version of firmware and the one immediately before it. Once I exit telnet, it goes away. 'Show buffers' indicated that there were 2 interpreter messages, also. I never remembered having this issue before and I use telnet all the time. I noticed it crippling my master the other night so much that it kept my shutdown timer from working.
  • tracktoystracktoys Junior Member Posts: 46
    FWIW, I tried to replicate the web server issue as mentioned in the original post, but had no issues. I was using Chrome.
  • DHawthorneDHawthorne Junior Member Posts: 4,584
    This is a very generic error, and nearly anything that eats a lot of processor ticks can cause it. For example, if you have a project with any number of the old Landmark-compatible DMS keypads, you will see a lot of it as they come online, settling down unless they are done. So the real question is, "is the web server the problem, or does it just use enough resources to cause something else to trigger the problem?" I have seen it caused by slow devices (like the DMS keypad) and by code that has to do a lot all at once (I had a routine that read an XML file that tended to cause it until I optimized it properly).

    And, for the record, it's not really a problem until the queue overflows and stops processing events altogether. The "maximum queue count" message itself is just a warning, and things might be bogging down at that point, but they are still working. When it tells you the queue has been suspended, than you are losing events.
  • a_riot42a_riot42 AMX Wizard Posts: 1,619
    DHawthorne wrote: »
    ne. So the real question is, "is the web server the problem, or does it just use enough resources to cause something else to trigger the problem?"

    I think its the webserver. A normally functioning system that has been running for weeks all of a sudden starts to print this to the console within seconds of pulling up the masters webpage.
    Paul
  • tracktoystracktoys Junior Member Posts: 46
    I've been thinking about this problem for the past couple of days. I have some issues with systems chewing up tons of CPU cycles. This by itself does not seem to be an issue, but when you open a telnet window in NS, it should not result in the master screeching to a halt.

    I ask a lot of my home system, processing no less than 6 RSS scrapes a few times an hour along with the normal device polling and system functions. I can understand that every so often when it's processing the bigger buffers and I'm doing something like updating my calendar and saving it to disk that it would slow up a bit. But why on earth is it this slow? I very rarely use define_program for anything, save for feedback on a couple of buttons in certain modules.

    I've noticed a few people here who are experiencing the similar issues. It could be that we are making some of the same mistakes in code, it could be that with each firmware upgrade more duet crap gets stuffed in there. Could it be that the hardware itself has run the course? I know that this is an apples to oranges comparison, but my home PC is a Core i7 quad with 6gb of memory. Given all of the tremendous strides in the chip market recently, is it time for AMX to get a better CPU with more memory under the hood to manage these systems and let them be more efficient? Is there any talk of updating the masters' hardware? Am I nuts?
  • Spire_JeffSpire_Jeff Formerly Caffeinated Programmer Posts: 1,917
    What is your Duet memory setting? I think the default is 3M, but running a couple of Duet modules and then opening the processor web server will cause very slow response times. Try going to 6M and see if that helps.

    Jeff
  • John GonzalesJohn Gonzales Junior Member Posts: 609
    This has worked for me everytime:
    http://www.amxforums.com/showpost.php?p=18466&postcount=11

    EDIT:
    Let me rephrase, the above works for dealing with the slow web console. I never checked to see if the message queues were overflowing during the slowness though.
    --John
  • a_riot42a_riot42 AMX Wizard Posts: 1,619
    tracktoys wrote: »
    Am I nuts?

    You might be but, I am seeing this on a new 3100 with almost nothing running on it. As soon as I surf to the masters webpage I get those CIpEvent::AddInternalEvent - Max Queue Count lines in diagnostics. They continue even after I have surfed away, and the master requires a reboot to get it to stop. I wish they would look into this.
    Paul
Sign In or Register to comment.