Lagging on Raids - Suspect Aura as Culprit

[spack]

Well, as the subject says, my SEQ graphical window lags/freezes/chunks-to-hell in big events. The program doesn't crash and eventually "breaks free" for a few seconds. This has only happened on big raid events on stuff like Vishimtar and Tunat. Looking at the terminal window I see a metric ton of Aura spell spam flying by. My best guess is that this new "aura" information is lagging out program or something. I'm sorry I'm not more technical in my description of the problem. Is anyone else seeing these problems? Other than on big events, everything seems to be working fine. And it's not even in all zones... we've raided some dragons in PoR and such and I"ve not seen it crap out on me.

[CeleSEQ]

I know purple is aware of the amount of traffic that the auras generate, but I'm not sure what if any changes might be made because of it. If that is the issue, the devs are aware... basically the auras generate a really vast amount of network traffic, and it's definitely possible that some machines might have trouble keeping up with the traffic from lots of auras all at once. Basically as I understand it, the auras are npcs which buff really rapidly and create lots of updates. Pretty poor protocol design really, they generate a disproportionate amount of network traffic.

There are other possibilities though. I have seen an issue with some versions of XFree86 where seq caused the X server to freak out and pin the CPU. I haven't seen that since RH changed to the xorg X server though, so there's possibly one thing you can look at.

If it is the auras we'll have to wait and see if someone can figure a way to make seq ignore some of the aura packets or otherwise deal with them more nicely.

[spack]

I have had issues on my laptop in previous versions of SEQ with X causing the CPU to pin out, as you said. It is Fedora Core 4, same as my desktop machine. However, observing SEQ chunking out tonight again, and running "top" command to look, it was SEQ that was eating up massive CPU, and not X. The best I can figure, it has to be the insane amount of traffic generated by the auras. My machine is a P4, 1.6GHz, with 512MB RAM. I know it's not super great, but it should be more than enough power and speed to run SEQ. I'm looking for suggestions here.

Edit: After actually watching this more closely, I have observed that all my information in the terminal window was literally lagging by several minutes after the fight was over. So, I'm getting so much information backed up for some reason, that the program is seems to be bottlenecked in processing all of it I guess. Help. What can I do?

[purple]

Maybe disable all the console messages (Interface->Terminal)? I don't know if it'll help. It's not like you can ignore auras with a pcap filter...

[Mano]

I don't think it is an issue specific to auras. This use to happen to me also a long time prior to PoR release. My setup at that time was...

P4 1.4ghz
640MB RAM
GeForce 4 ti4200 128MB
40GB HD
8GB HD

I had Windows 2k3 Server installed on the P4
I had Suse Linux installed via VMWare 4
I had 256MB Ram allocated to the linux virtual machine
I used the 8GB HD for the linux virtual machine.

With this setup I had to use an ARQ value of 1024 to prevent seq from crashing (prior to learning of the kernel settings you can change). Every now and then SEQ would cause this processing overload.

Now, I'm using the same hardware with only a linux install and run with an ARQ of 256 and the kernel changes outlined in another post. I have not had the cpu issue since.

Due to the way auras affect network traffic it could make it appear to be an aura problem but I would suspect it is associated with buffered packets in some way? Is that a viable option?

[uRit1u2CBBA=]

One thing that was found in my guild to work on the "lag times when groups get shuffled" is to find the DBG.TXT file in the eq\logs folder and make it READ-ONLY. It gets hammered with messages whenever groups got moved around with arua messages.

[spack]

I figured since I started this thread, I should post an update with some pertinent information. After having this bizare issue for a while, I found that if I logged out of my Fedora Core 4 account (not rebooting, just logging out) and then logged back in, I woudn't lag out. But after a day or so, I'd have to make sure I logged out and back in. I'd do this every day before raid just to make sure. My whole Gnome desktop would be lagging. The best I can figure is that there must have been some kind of resource losses with X. Celeseq mentioned something about this.

Since then I have installed SuSE 10.1 on my machine. This machine is a slightly older Dell machine. SuSE did not like the network card in it, which was some Davicom card that had notible problems I found out. Simple solution, I slapped in a Linksys 100LNTX (did I get that model right? going by memory here) card in it and it's working beautiful. Only been running this a little over a week now. Haven't had any issues yet. I'm going to chock this up to a FUCKDORA CORE 4 issue. While it is not my intention to start a distro flame war here, please believe me when I say that between 4 different machines, and Fedora Core 3 and 4 on all of them, I have had several very bizarre problems. Since SuSE has deployed opensuse.org and actually has dowloadable ISO files for CD/DVD images now (they didn't used to), I've tossed my FC 3/4 disks and am done with that crap distro.

P.S. Again, I have no interest in starting a distro flame war here. Your mileage may vary. I have many reasons for using SuSE Linux that don't need to be listed here.

[seqfan]

I'm running FC4 on worse hardware than you. ShowEQ runs great. I rarely have to reboot my fc4 box. It runs for weeks with SEQ running the whole time. As a matter of fact, it would probably run for months if I'd fork out cash for a new UPS (battery is crapping out of the current one) or if my local Electric Cooperative would be reliable. I think you answered your own question in that it was partially due to the crappy NIC you had in it. I'm betting that if you load fc4 on that suse machine it would run just as well.