Re: Network Slowdowns?
Matthew Dillon wrote: Typically when that happens you get 'watchdog timeout' messages on the console. Ehm, watchdog timeouts kill the network completely, don't they? I got a couple of them recently, but I'm pretty sure those ones weren't driver-related, cause in another place it runs fine now (and filters a whole Gbit network). All killed the network completely, had to reboot.
Re: Network Slowdowns?
In [EMAIL PROTECTED], Bill Hacker [EMAIL PROTECTED] mentions: Jamie wrote: Curious if anyone else has experienced this: *snip* (details) Looks suspiciously like a DoS attack Referring to the freeze ups? OR a service run amok - What is your firewall logging? Tons of connections going in and out just before the lockup periods. (way more than I could actually read the log for. Unfortunately, I can't seem to duplicate it. Although, reducing the number of connections did make the problem go away. - what is 'top' showing? Load averags, top, etc.. had all shown normal activity, even when the network was frozen. It's sounding to me like these are two separate things though. Jamie -- http://www.geniegate.comCustom web programming [EMAIL PROTECTED] (rot13)User Management Solutions
Re: Network Slowdowns?
In [EMAIL PROTECTED], Matthew Dillon [EMAIL PROTECTED] mentions: There are a couple of possibilities. One is that the driver is broken or can't handle that particular card or motherboard device. Typically when that happens you get 'watchdog timeout' messages on the console. Run 'dmesg' to see if you get any such messages. Nope, no watchdog messages. I do get several of these though: xl0: transmission error: 90 xl0: tx underrun, increasing tx start threshold to 120 bytes I read some place this had something to do with DMA buffers and was more or less harmless. It's a 3com card: xl0: 3Com 3cSOHO100-TX OfficeConnect port 0x6800-0x687f mem 0xe800-0xe87f irq 10 at device 11.0 on pci0 On the other end, Linux calls them this: 3Com PCI 3cSOHO100-TX Hurricane (I use 3Com on all machines) I do get an awful lot of these: nfs: server mercury not responding, still trying (mercury is the dragonflybsd machine) of course during these times, the nfs mounted directories and all programs depending on them are unusable. I can usually trigger this by just writing a lot of data to the NFS directory. I don't know if it's related or not. If you are not getting watchdog timeout messages then it could be cabling or auto-negotiation problem, or an ARP problem (two machines on the network trying to use the same IP address). Hmm.. I really doubt it's a problem with two machine attempting to use the same IP, I use static IP's below 192.168.1.100. I'll check some more into this to be sure. The way it seems to happen is it gets progressively slower until I reset the card. Repeat.. What kernel version are you running? -Matt 1.6.1-RELEASE DragonFly 1.6.1-RELEASE #4: Thu Sep 14 09:41:54 CDT 2006 Someone else suggests it could be a duplex mismatch, I'm looking in to that at the moment, (I have never messed around with that, left them at their defaults.) Jamie -- http://www.geniegate.comCustom web programming [EMAIL PROTECTED] (rot13)User Management Solutions
Re: Network Slowdowns?
Jamie wrote: *SNIP* (all details already posted) (I use 3Com on all machines) Without digressing into decades of *why*, I can just about guarantee that replacing the offending card with almost-anything-else, from el-cheapo Realtek to Gig-E Intel, probable exception of anything-SiS, will cure the problem without further ado. Cost you perhaps US$ 10 to try that if you don't already have some other make of NIC handy. 3Com seem to fall into either 'solved all problems' or 'created a problem' extremes all too often. We *always* replace 3Com on general principal when encountered, and at our own (not client) expense. Not about right or wrong, its about what works *always* and what doesn't always work. The time saved the past dozen years has been more than worth the very modest cost of replacement NIC's. Life is too short etc. JM2CW Bill
Re: Network Slowdowns?
On 07 Oct 2006 07:39:21 GMT, Jamie [EMAIL PROTECTED] wrote: In [EMAIL PROTECTED], Matthew Dillon [EMAIL PROTECTED] mentions: There are a couple of possibilities. One is that the driver is broken or can't handle that particular card or motherboard device. Typically when that happens you get 'watchdog timeout' messages on the console. Run 'dmesg' to see if you get any such messages. Nope, no watchdog messages. I do get several of these though: xl0: transmission error: 90 xl0: tx underrun, increasing tx start threshold to 120 bytes From 3com's DS, 0x90 means your host's feeding packets rate is too low. How many times you have seen this error? Let's take it as N line 433 in /usr/src/sys/dev/netif/xl/if_xlreg.h #define XL_MIN_FRAMELEN 60 Say N == 3, then change above line to (3 + 1) * 60, i.e. #define XL_MIN_FRAMELEN 240 But keep it less then 1540 See whether the problem still plagues you Best Regards, sephe -- Live Free or Die
Re: Network Slowdowns?
Maybe the tx start threshold should be increased to 240 per default for the xl driver. I get this log message for both my two xl cards (internal network and vdsl network) and it always stops at 240. -Jonas On 10/7/06, Sepherosa Ziehau [EMAIL PROTECTED] wrote: On 07 Oct 2006 07:39:21 GMT, Jamie [EMAIL PROTECTED] wrote: In [EMAIL PROTECTED], Matthew Dillon [EMAIL PROTECTED] mentions: There are a couple of possibilities. One is that the driver is broken or can't handle that particular card or motherboard device. Typically when that happens you get 'watchdog timeout' messages on the console. Run 'dmesg' to see if you get any such messages. Nope, no watchdog messages. I do get several of these though: xl0: transmission error: 90 xl0: tx underrun, increasing tx start threshold to 120 bytes From 3com's DS, 0x90 means your host's feeding packets rate is too low. How many times you have seen this error? Let's take it as N line 433 in /usr/src/sys/dev/netif/xl/if_xlreg.h #define XL_MIN_FRAMELEN 60 Say N == 3, then change above line to (3 + 1) * 60, i.e. #define XL_MIN_FRAMELEN 240 But keep it less then 1540 See whether the problem still plagues you Best Regards, sephe -- Live Free or Die
Re: Can't start KDE or TWM with KDM
Le Fri, 06 Oct 2006 17:13:57 +0200, walt [EMAIL PROTECTED] a écrit: Thanks Thomas and Walt.
Runtime missing-symbol errors?
I'm trying to figure out why evolution-2.8 doesn't work on DF: (evolution-2.8:39636): evolution-shell-WARNING **: Cannot activate 'OAFIID:GNOME_Evolution_Mail_Component:2.8': Can't find symbol Bonobo_Plugin_info in `/usr/pkg/lib/evolution/2.8/components/libevolution-mail.so' Note that the error does not come from ld-elf.so trying to load a stale library -- this message comes from evolution itself. Does this imply the use of dlopen(), maybe? And here is part B of the puzzle: #nm /usr/pkg/lib/evolution/2.8/components/libevolution-mail.so | grep Bonobo_Plugin 000aeb80 D Bonobo_Plugin_info So the 'missing' symbol really is defined in the library, but it can't be found for some reason. Any ideas what is going wrong?
KDE and SSL still not working
Ive just recompiled kdelibs3 after a couple of months but find out that SSL in KDE is still broken. Is it moving anywhere closer to be fixed? Petr