Re: Mbuf Clusters on 4.8
John Bdckstrand wrote: Ive been googling quite a bit now for problems with running out of mbuf clusters. Im basically sending a 30k datachunk down 1000-4000 connections, but 1000 is more than enough to quickly fill upp 8192 mbuf clusters. I also tried setting maximum amount of mbuf clusters to 65536, but that only made the box hard-wire 86MB of 96MB RAM, making it just as unsuable as a dead machine. It isn't the amount data you are sending but the overhead required for each network connection. I would recommend adding more RAM. Bump it up to 256MB. With mbuf set at 65536 and using 86MB, you should still have plenty left for your application. If that doesn't solve your problem or you can't add more memory, then you'll want to look at controling the number of simultaneous connections to a number that your box can handle. Of course, when the machine runs out of mbuf clusters, it dies. I also found this with google: Finally, the fact that FreeBSD 3.x panics when it runs out of mbuf clusters is a well-known problem. The solution is to not let it run out of mbuf clusters by configuring a sufficient number for them. From this it sounds as it is a problem that should be fixed, but it obviously isnt in 4.8. Is this behaviour now considered acceptable? And if so, doesnt this make FreeBSD extremely easy to kill using a simple DOS-attack? Is this fixed in any way on 5.1? Yup, that is what DoS attack is... exhaustion of one or more resources of the victim. P2P software is an easy way to exhaust mbuf buffers on a box. P2P software(e.g. edonkey) can be a useful network stress tool; opens lots of connections and pushes a lot of data. My experience with mbuf exhaustion on a 4-stable boxes has been the box basically loses network connectivty until it can recover some buffers. The box is still responsive from the console and killing the offending application from the console will free up the mbufs and restore network connectivity. good luck, greg ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Mbuf Clusters on 4.8
On Thu, Jun 26, 2003 at 03:03:20AM +0200, John B?ckstrand wrote: Ive been googling quite a bit now for problems with running out of mbuf clusters. Im basically sending a 30k datachunk down 1000-4000 connections, but 1000 is more than enough to quickly fill upp 8192 mbuf clusters. I also tried setting maximum amount of mbuf clusters to 65536, but that only made the box hard-wire 86MB of 96MB RAM, making it just as unsuable as a dead machine. Of course, when the machine runs out of mbuf clusters, it dies. I also found this with google: Finally, the fact that FreeBSD 3.x panics when it runs out of mbuf clusters is a well-known problem. The solution is to not let it run out of mbuf clusters by configuring a sufficient number for them. From this it sounds as it is a problem that should be fixed, but it obviously isnt in 4.8. Is this behaviour now considered acceptable? And if so, doesnt this make FreeBSD extremely easy to kill using a simple DOS-attack? Is this fixed in any way on 5.1? --- John B?ckstrand It's not panicking, it's running out of resources. Whenever you have this sort of problem you need to provide more information, there is absolutely no way I can help you like this. You need to, at the very minimum, give us 'netstat -m' output and make a serious attempt at figuring out what is consuming so many clusters. You could be running out of clusters but you could also be running out of memory before you run out of clusters, in which case you should probably _not_ increase nmbclusters and instead fix the underlying problem instead (re-work your application). In such a scenario, blindly bumping up nmbclusters can make the problem worse. Even if you had 'unlimited' (or dynamically growing) nmbclusters, you'd _still_ have the same problem and, what's more, it could actually render your system even more unusable as the machine would not be able to allocate memory for other more important uses. Yes, you are right, I didnt get a panick. Firstly, heres is /var/log/messages from when the box hung: Jun 26 02:29:30 sandbsd /kernel: All mbuf clusters exhausted, please see tuning(7). Jun 26 02:29:35 sandbsd last message repeated 4 times Jun 26 02:29:35 sandbsd /kernel: rl0: watchdog timeout Jun 26 02:29:36 sandbsd /kernel: All mbuf clusters exhausted, please see tuning(7). Jun 26 02:29:56 sandbsd last message repeated 17 times Jun 26 02:29:57 sandbsd /kernel: rl0: watchdog timeout I cant give netstat -m after it hung, obviously, but with 8192 max mbuf clusters I can see some time before the hang that peak is at 8192 clusters, I also saw lots of requests for memory denied, and requests for memory delayed, but no calls to drain routines. I also saw about 20MB allocated to network (60% of mb_map). This was the last time I tried hanging it, and I somehow thought it mas managing. After all, denying requests means it did actually cope with a memory starvation. The NIC is a realtek 8139. I wonder what significance those watchdog timeout messages have? I might try with a 3com later today (any card which doesnt require mbufs should be finer, if there is any such thing). --- John Bäckstrand ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Mbuf Clusters on 4.8
From this it sounds as it is a problem that should be fixed, but it obviously isnt in 4.8. Is this behaviour now considered acceptable? And if so, doesnt this make FreeBSD extremely easy to kill using a simple DOS-attack? Is this fixed in any way on 5.1? Yup, that is what DoS attack is... exhaustion of one or more resources of the victim. P2P software is an easy way to exhaust mbuf buffers on a box. P2P software(e.g. edonkey) can be a useful network stress tool; opens lots of connections and pushes a lot of data. My experience with mbuf exhaustion on a 4-stable boxes has been the box basically loses network connectivty until it can recover some buffers. The box is still responsive from the console and killing the offending application from the console will free up the mbufs and restore network connectivity. Ah, unfortunately my box doesnt respond even to keyboard events (caps lock etc). The behaviour you describe I find totally acceptable on the other hand. And the software Im writing happens to be p2p-related, but its not a edonkey server. :) --- John Bäckstrand ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Is there something sppecial about pass4 ?
Machine is 4.8-STABLE smaug# camcontrol devlist COMPAQ BD009122BA 3B07 at scbus0 target 0 lun 0 (pass0,da0) COMPAQ BD009122C6 B016 at scbus0 target 1 lun 0 (pass1,da1) COMPAQ BD009122BA 3B07 at scbus0 target 2 lun 0 (pass2,da2) COMPAQ BD00962373 BCJE at scbus0 target 3 lun 0 (pass3,da3) IBM IC35L018UCPR15-0 S70Hat scbus0 target 5 lun 0 (pass4,da4) smaug# ls -l /dev/pass* crw--- 1 root operator 31, 0 Jun 26 15:54 /dev/pass0 crw--- 1 root operator 31, 1 Jun 26 15:54 /dev/pass1 crw--- 1 root operator 31, 2 Jun 26 15:54 /dev/pass2 crw--- 1 root operator 31, 3 Jun 26 15:54 /dev/pass3 No pass4. I cant make pass4 with MAKEDEV either - plus I note that in the MAKEDEV scrip then 'all' explicitly does 'sh MAKEDEV pass4 xpt2' using mknod by hand seems to work - but I cant see why MAKEDEV wont do it. -pcf. ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Is there something sppecial about pass4 ?
On Thu, Jun 26, 2003 at 15:59:33 +0100, Pete French wrote: Machine is 4.8-STABLE smaug# camcontrol devlist COMPAQ BD009122BA 3B07 at scbus0 target 0 lun 0 (pass0,da0) COMPAQ BD009122C6 B016 at scbus0 target 1 lun 0 (pass1,da1) COMPAQ BD009122BA 3B07 at scbus0 target 2 lun 0 (pass2,da2) COMPAQ BD00962373 BCJE at scbus0 target 3 lun 0 (pass3,da3) IBM IC35L018UCPR15-0 S70Hat scbus0 target 5 lun 0 (pass4,da4) smaug# ls -l /dev/pass* crw--- 1 root operator 31, 0 Jun 26 15:54 /dev/pass0 crw--- 1 root operator 31, 1 Jun 26 15:54 /dev/pass1 crw--- 1 root operator 31, 2 Jun 26 15:54 /dev/pass2 crw--- 1 root operator 31, 3 Jun 26 15:54 /dev/pass3 No pass4. I cant make pass4 with MAKEDEV either - plus I note that in the MAKEDEV scrip then 'all' explicitly does 'sh MAKEDEV pass4 xpt2' using mknod by hand seems to work - but I cant see why MAKEDEV wont do it. The number for the passthrough device is actually how many to make. So when you do a MAKEDEV pass4, you get pass0-3. So try: sh MAKEDEV pass8 You should wind up with pass0-pass7. Ken -- Kenneth Merry [EMAIL PROTECTED] ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Can I remove lib-elf.so.1?
On Wed, 25 Jun 2003, Geoffrey T. Falk wrote: GTF I am cleaning up my 4-STABLE system. After a fresh installworld, I am GTF looking at files that did not get touched by the install. Is it safe to GTF remove all such files? GTF GTF In particular, I am looking at /usr/libexec/lib-elf.so.1, which has the GTF schg flag set. I am afraid to remove this for fear of making my system GTF unbootable. Could somebody please reassure me that it is ok to noschg GTF and remove this old file? My 4.8-R and 4.8-S systems do not have this file: [EMAIL PROTECTED]:~ l -o /usr/libexec/lib-elf.so.1 ls: /usr/libexec/lib-elf.so.1: No such file or directory [EMAIL PROTECTED]:~ uname -a FreeBSD woozle.rinet.ru 4.8-STABLE FreeBSD 4.8-STABLE #4: Wed Apr 16 17:28:12 MSD 2003 [EMAIL PROTECTED]:/usr/obj/lh/src/sys/woozle i386 [EMAIL PROTECTED]:/var# l -o /usr/libexec/lib-elf.so.1 ls: /usr/libexec/lib-elf.so.1: No such file or directory [EMAIL PROTECTED]:/var# uname -a FreeBSD kucha.rinet.ru 4.8-RELEASE FreeBSD 4.8-RELEASE #2: Tue Apr 1 21:02:29 MSD 2003 [EMAIL PROTECTED]:/usr/obj/FreeBSD/src.stable-48/sys/kucha i386 So, I suppose this file can be safely removed. Sincerely, D.Marck [DM5020, MCK-RIPE, DM3-RIPN] *** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [EMAIL PROTECTED] *** ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Can I remove lib-elf.so.1?
On Wed, Jun 25, 2003 at 11:52:55PM -0600, Geoffrey T. Falk wrote: I am cleaning up my 4-STABLE system. After a fresh installworld, I am looking at files that did not get touched by the install. Is it safe to remove all such files? No, it is not always safe. Some files are only updated if the new file is different from the old file. The include files are an example of this as to avoid fooling makefiles into believing that stuff need to be recompiled when it is just the date of an include file that has changed and not its contents. In particular, I am looking at /usr/libexec/lib-elf.so.1, which has the schg flag set. I am afraid to remove this for fear of making my system unbootable. Could somebody please reassure me that it is ok to noschg and remove this old file? There shouldn't be any file named /usr/libexec/lib-elf.so.1 so if you actually have one it is safe to remove it (and time to start investigating where it came from), but if you actually meant /usr/libexec/ld-elf.so.1 then you should *NOT* remove it. /usr/libexec/ld-elf.so.1 is needed for running any dynamically linked programs. It is not quite as critical as /sbin/init or /kernel but nearly so. It is precisely because this file is so critical that it is not updated unless it has actually changed. In general the few files that have the schg flag set has it for good reasons. -- Insert your favourite quote here. Erik Trulsson [EMAIL PROTECTED] ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Virus Alert
*** THIS IS AN AUTOMATED MESSAGE *** A mail message sent from this account to a Cisco employee had an attachment called your_details.zip. The file your_details.zip contained a virus. Please contact your local System Administrator for further assistance. ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
apache panics on a recent 4.8-STABLE
Yesterday I begin a couple of update to the latest 4.8-STABLE. After that the two boxes continues to go in panics as soon as Apache (1.3 from the ports, also freshly recompiled, 2.0.x seems NOT to hang) starts. I don't know if it is related to the other thread : Kernel core dump in recent 4.8-STABLE but it is easily reproducible by cvsupping to a today -STABLE and then running apache 1.3 ... Thanks very much for any info you con provide. Best Regards, Gianmarco Giovannelli , Unix expert since yesterday http://www.gufi.org/~gmarco ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: garbled top display
Eric J. Chet wrote: [...] Just a sanity check. I'm seeing a garbled display with top. I rebuild kernel and world a couple times, I even rm -rf /usr/include to make sure I was up to date. Anybody else seeing this? Define garbled. Thanks, David Sure Script started on Mon Jun 23 10:39:06 2003 bash-2.05a$ top [H[2Jlast pid: 3359; load averages: 0.06, 0.02, 0.00[1;56H up 1+13:20:42[1;72H10:39:09 10 processes: 1 running CPU states: % user, % nice, % system, % interrupt, % idle Mem: 16M Active, 55M Inact, 26M Wired, 8748K Cache, 22M Buf, 15M Free Swap: 256M Total, 52K Used, 256M Free PID USERNAME PRI NICE SIZERES STATETIME WCPUCPU COMMAND 1 '++*00'* -22 -52 0K 0K ? -128??? 0.15% 0.15% 968972 root -22 -52 3083M 0K ?6??? 0.00% 0.05% [... etc etc ...] wow. I've seen garbled top displays before, like the RES going -ve or things like that. But never anything like this. I am reminded of the scene in Poltergeist, where the couple get a paranormal expert into the house to take a look, and he natters on about how he saw a toy car roll across a room over the space of six hours... and then they open the door... It looks to me like your structs are out of whack, and so when top comes along to read an element of a struct, it's using the wrong offset from base, and hence printf (or whatever) renders garbage. I.e., as if different headers are used in compiling the kernel and the rest. Does ps give reasonable output? Are you sure you really cleaned out /usr/src before recompiling? Has the compiler been upgraded recently? David ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: IBM ServeRaid - ips driver?
I have a brand new IBM 335 with IDE controller and I have the same keyboard problem. The keyboard is working at bios/bootloader and stop to work when freebsd kernel boot. I get this error atkbd: unable to set the command byte. Any clue? uname -a FreeBSD 5.1-RELEASE FreeBSD 5.1-RELEASE #0: Thu Jun 5 02:55:42 GMT 2003 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/GENERIC i386 Jun 26 09:46:10 kernel: atkbdc0: Keyboard controller (i8042) port 0x60,0x64 irq 1 on acpi0 Jun 26 09:46:10 kernel: atkbd0: AT Keyboard flags 0x1 irq 1 on atkbdc0 Jun 26 09:46:10 kernel: atkbd: unable to set the command byte. Jun 26 09:46:10 kernel: device_probe_and_attach: atkbd0 attach returned 6 Jun 26 09:46:10 kernel: psm0: unable to set the command byte. On Mon, 2 Jun 2003, Arnvid Karstad wrote: Hiya all, Just saw that the 5.1 RC/Beta had finally support ips/serveraid controllers. (Even tho 4Lx and 5i seem to have LSI chipsets contra the older ones) But when trying to boot the 5.1RC1 mini iso it failes to detect my keyboard... The server is an IBM xSeries 335 1 u server with and onboard LSI (mpt0) kontroller for the boot disks and an 4Lx used for EXP300 external storage. Try a different keyboard mfr. or a USB keyboard. -- Dominic Pageau ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: IBM ServeRaid - ips driver?
verbose boot Jun 26 14:50:59 kernel: atkbdc0: Keyboard controller (i8042) port 0x60,0x64 irq 1 on acpi0 Jun 26 14:50:59 kernel: atkbd0: AT Keyboard flags 0x1 irq 1 on atkbdc0 Jun 26 14:50:59 kernel: atkbd: the current kbd controller command byte 0065 Jun 26 14:50:59 kernel: atkbd: unable to set the command byte. Jun 26 14:50:59 kernel: device_probe_and_attach: atkbd0 attach returned 6 Jun 26 14:50:59 kernel: psm0: unable to allocate IRQ Jun 26 14:50:59 kernel: psmcpnp0 irq 12 on acpi0 Jun 26 14:50:59 kernel: psm0: current command byte:0065 Jun 26 14:50:59 kernel: psm0: unable to set the command byte. On Mon, 2 Jun 2003, Arnvid Karstad wrote: Hiya all, Just saw that the 5.1 RC/Beta had finally support ips/serveraid controllers. (Even tho 4Lx and 5i seem to have LSI chipsets contra the older ones) But when trying to boot the 5.1RC1 mini iso it failes to detect my keyboard... The server is an IBM xSeries 335 1 u server with and onboard LSI (mpt0) kontroller for the boot disks and an 4Lx used for EXP300 external storage. Try a different keyboard mfr. or a USB keyboard. -- Dominic Pageau ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]