cvs kills connections

[Stefan Wollny]

Oh dear ... cvs w/ OpenBSD-amd64-current is a PITA at present.

I describe my setup, the issue I noticed and what I tried and provide
some informmation like dmesg at the end.

I have this IBM/Lenovo T60 running OpenBSD's amd64-current:
OpenBSD 5.6-current (GENERIC.MP) #413: Sat Oct 11 15:59:03 MDT 2014
dera...@amd64.openbsd.org:/usr/src/sys/arch/amd64/compile/GENERIC.MP

(But the issue is hunting me for the least two weeks now. The system has
been set up from scratch - I have the last dmesg's as well...)

On misc@ I found two hints that show up in the information provided at
the end: I have set 'net.inet.ip.ifq.maxlen' to 1024 and
added 'set optimization conservative' to pf.conf.

I use 'adsuck', so resolv.conf lives in /var/adsuck/files while
/etc/resolv.conf only has the one line
nameservers 127.0.0.1

(And YES - the issue arises as well without adsuck!)

No matter if I connect to the router via dhcp (either dhcp up in
hostname.em0 or trunk0) or if I use an outright hostname.em0
(inet 192.168.178.31 255.255.255.0 NONE) after a few seconds trying to
GET or to UPdate cvs the connection is lost.

I use this system as my day2day-workhorse with a Fluxbox-based desktop.
Firefox, Thunderbird, libreoffice, gimp. Nothing fancy,
nothing to bother as the issue comes up already the console-level: The
moment I run 'cvs -q up -Pd' the system looses it's connection and I get
the note
ssh: connect to host ftp.hostserver.de port 22: No route to host
 cvs [update aborted]: end of file from server (consult above messages
if any)

 or
/usr/ports $ sudo cvs -q up -Pd
 cannot create_adm_p /tmp/cvs-serv32110/net/py-flowtools/pkg
 No space left on device

No space left on device - WTF???

~ $ df -h /
Filesystem SizeUsed   Avail Capacity  Mounted on
/dev/sd0a  4.9G   66.9M4.6G 1%/
~ $ df -h /tmp
Filesystem SizeUsed   Avail Capacity  Mounted on
/dev/sd0d  4.9G   18.0K4.7G 0%/tmp
~ $ df -h /var
Filesystem SizeUsed   Avail Capacity  Mounted on
/dev/sd0e  9.8G258M9.1G 3%/var
~ $ df -h /usr
Filesystem SizeUsed   Avail Capacity  Mounted on
/dev/sd0f 29.5G8.4G   19.6G30%/usr
~ $ df -h /home
Filesystem SizeUsed   Avail Capacity  Mounted on
/dev/sd0g  397G   45.4G332G12%/home

pfctl -s states shows plenty of
all tcp 192.168.178.31:17141 - 217.31.80.35:80   TIME_WAIT:TIME_WAIT

(whois 217.31.80.35 tells that it is actually the mirror-site
ftp.hostserver.de)

routes for the default to gateway go up from 68 after start to 38700 at
the moment the connection is lost ('route show' at the very end of this
mail).

The hardware is unchanged, the network is doing fine with a windooze-box.

Below I provide the dmesg, pf.conf, output of sysctl -a, pfctl -s
all and route show from right after startup and from the moment the
connection is lost. (I am lost as well ... :-( in particular I'd like to
have my ports-tree up2date for ~current)

Can anybody help me, please?

TIA!

Cheers,
STEFAN




#
#
DMESG
#
#

OpenBSD 5.6-current (GENERIC.MP) #413: Sat Oct 11 15:59:03 MDT 2014
dera...@amd64.openbsd.org:/usr/src/sys/arch/amd64/compile/GENERIC.MP
real mem = 3203203072 (3054MB)
avail mem = 3109289984 (2965MB)
mpath0 at root
scsibus0 at mpath0: 256 targets
mainbus0 at root
bios0 at mainbus0: SMBIOS rev. 2.4 @ 0xe0010 (68 entries)
bios0: vendor LENOVO version 79ETC9WW (2.09 ) date 12/22/2006
bios0: LENOVO 2007VG2
acpi0 at bios0: rev 2
acpi0: sleep states S0 S3 S4 S5
acpi0: tables DSDT FACP SSDT ECDT TCPA APIC MCFG HPET SLIC BOOT SSDT
SSDT SSDT SSDT
acpi0: wakeup devices LID_(S3) SLPB(S3) LURT(S3) DURT(S3) EXP0(S4)
EXP1(S4) EXP2(S4) EXP3(S4) PCI1(S4) USB0(S3) USB1(S3) USB2(S3) USB7(S3)
HDEF(S4)
acpitimer0 at acpi0: 3579545 Hz, 24 bits
acpiec0 at acpi0
acpimadt0 at acpi0 addr 0xfee0: PC-AT compat
cpu0 at mainbus0: apid 0 (boot processor)
cpu0: Intel(R) Core(TM)2 CPU T7200 @ 2.00GHz, 1994.60 MHz
cpu0:
FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CFLUSH,DS,ACPI,MMX,FXSR,SSE,SSE2,SS,HTT,TM,PBE,SSE3,DTES64,MWAIT,DS-CPL,VMX,EST,TM2,SSSE3,CX16,xTPR,PDCM,NXE,LONG,LAHF,PERF
cpu0: 4MB 64b/line 16-way L2 cache
cpu0: smt 0, core 0, package 0
mtrr: Pentium Pro MTRR support, 8 var ranges, 88 fixed ranges
cpu0: apic clock running at 166MHz
cpu1 at mainbus0: apid 1 (application processor)
cpu1: Intel(R) Core(TM)2 CPU T7200 @ 2.00GHz, 1994.33 MHz
cpu1:
FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CFLUSH,DS,ACPI,MMX,FXSR,SSE,SSE2,SS,HTT,TM,PBE,SSE3,DTES64,MWAIT,DS-CPL,VMX,EST,TM2,SSSE3,CX16,xTPR,PDCM,NXE,LONG,LAHF,PERF
cpu1: 4MB 64b/line 16-way L2 cache
cpu1: smt 0, core 1, package 0
ioapic0 at mainbus0: apid 1 pa 0xfec0, version 20, 24 pins
ioapic0: misconfigured as apic 2, remapped 

Re: cvs kills connections

[Nick Holland]

On 10/12/14 07:43, Stefan Wollny wrote:
...
  The
 moment I run 'cvs -q up -Pd' the system looses it's connection and I get
 the note
 ssh: connect to host ftp.hostserver.de port 22: No route to host
  cvs [update aborted]: end of file from server (consult above messages
 if any)

that's pretty clear, isn't it?  The SSH connection is failing.
Troubleshoot it at that level, or even traceroute.  Don't complain about
CVS when the SSH transport is failing...and not sure I'd talk about SSH
when you have a no route to host message.

You've got some funky stuff in your pf.conf, including apparently
blocking your local subnet (which appears to be 192.168.178.0/??,
blocked with your broken table).  Shouldn't be a problem, but not sure
what your firewall/router and/or DHCP server will do with a machine that
doesn't appear to be there.

  or
 /usr/ports $ sudo cvs -q up -Pd
  cannot create_adm_p /tmp/cvs-serv32110/net/py-flowtools/pkg
  No space left on device
 
 No space left on device - WTF???

almost always means ...on the CVS server.  cvs has a nasty habit of
leaving trash laying around in the /tmp directory if someone aborts a
connection, and by trash, I mean hundreds of MB sometimes.  Also...cvs
is one of the few apps I've seen burn inodes faster than disk space --
if the anoncvs tmp directory is its own partition, it generally needs
more-than-normal number of inodes, and a nightly erasure of anything
older than say 24 hours.  Notify the anoncvs mirror administrator.

Nick.

Re: cvs kills connections

[Stefan Wollny]

Am 10/12/14 um 16:28 schrieb Nick Holland:
 On 10/12/14 07:43, Stefan Wollny wrote: ...
 The moment I run 'cvs -q up -Pd' the system looses it's
 connection and I get the note ssh: connect to host
 ftp.hostserver.de port 22: No route to host cvs [update aborted]:
 end of file from server (consult above messages if any)

 that's pretty clear, isn't it?  The SSH connection is failing.
 Troubleshoot it at that level, or even traceroute.  Don't complain
 about CVS when the SSH transport is failing...and not sure I'd talk
 about SSH when you have a no route to host message.


Hi Nick,

thanks for replying.

My guess on cvs is due to only when trying to update /usr/src,
/usr/ports and /usr/xenocara. Even though I usually use
ftp.hostserver.de for being the closest I have tried with others as
well: openbsd.cs.fau.de, ftp.bytemine.net and ftp.openbsd.org.

Moreover I can ssh into any other machine in my net without trouble -
even though I have to admit that I did so only to one
~current/i386-machine.

But of course I am interested in following your thoughts: What can I
do to investigate further? Or is this s.th. libressl-related that will
be cured in a few days with some of the next snapshots?


 You've got some funky stuff in your pf.conf, including apparently
 blocking your local subnet (which appears to be 192.168.178.0/??,
 blocked with your broken table).  Shouldn't be a problem, but not
 sure what your firewall/router and/or DHCP server will do with a
 machine that doesn't appear to be there.

I did not deliberately block my subnet, of course, as my subnet is not
part of the tabel blockes_zones (but 192.168.0.0 is).
But could this be a result fromn the following rule:
block in quick from { $broken urpf-failed no-route } to any

The moment the route is lost the block-rules matches - correct?

Otherwise may I ask you to give me an other clue on where my rules are
'suboptimal'?


 or /usr/ports $ sudo cvs -q up -Pd cannot create_adm_p
 /tmp/cvs-serv32110/net/py-flowtools/pkg No space left on device

 No space left on device - WTF???

 almost always means ...on the CVS server.  cvs has a nasty habit
 of leaving trash laying around in the /tmp directory if someone
 aborts a connection, and by trash, I mean hundreds of MB
 sometimes.  Also...cvs is one of the few apps I've seen burn inodes
 faster than disk space -- if the anoncvs tmp directory is its own
 partition, it generally needs more-than-normal number of inodes,
 and a nightly erasure of anything older than say 24 hours.  Notify
 the anoncvs mirror administrator.

 Nick.

Thanks - I will notify the admins, Sebastian Benoit and Florian Obser.

Cheers,
STEFAN