named vs 2.5.64-mm5
Greetings -
2.5.64-mm4 and -mm5 seem more rugged than previous kernels, but there are a couple of minor nits - one of them is the tendency of named (which appears to work reliably under 2.4) to go catatonic under recent 2.5.6x kernels -
More verbose kernel logging may shed some light - or is this just a red herring? I get a tons of these in 2.5.64-mm5:
<...>
process named' is using obsolete setsockopt SO_BSDCOMPAT process named' is using obsolete setsockopt SO_BSDCOMPAT process `named' is using obsolete setsockopt SO_BSDCOMPAT <...>
Anybody here running a compliant version of named?
(This is the bind 9.2.1 which ships with Red Hat 8.0)
Best Regards,
Joe
Quoted 8 lines · jjs
Greetings -
2.5.64-mm4 and -mm5 seem more rugged than previous kernels, but there are a couple of minor nits - one of them is the tendency of named (which appears to work reliably under 2.4) to go catatonic under recent 2.5.6x kernels -
Quoted 9 lines
More verbose kernel logging may shed some light - or is this just a red herring? I get a tons of these in 2.5.64-mm5:
<...>
processnamed' is using obsolete setsockopt SO_BSDCOMPAT processnamed' is using obsolete setsockoptSO_BSDCOMPATprocess `named' is using obsolete setsockoptSO_BSDCOMPAT<...>
The changelog has:
# --------------------------------------------
# 03/03/08 [email protected] 1.1083
# [NET]: Nuke SO_BSDCOMPAT.
# --------------------------------------------
Maybe James can tell us what is going on here.
We should at least place a cap on the number of times that message is printed.
Quoted 11 lines · Andrew Morton
The changelog has:
# --------------------------------------------
# 03/03/08 [email protected] 1.1083
# [NET]: NukeSO_BSDCOMPAT.
# --------------------------------------------Maybe James can tell us what is going on here.
We should at least place a cap on the number of times that message is printed.
Feel free to send a patch for that.
SO_BSDCOMPAT has had ZERO side effects since 2.0.x, and it's been thus scheduled to be removed for years. It was merely a binary state passed in and out of the kernel to the user and had no effect on socket behavior at all.
Any application still referencing this ancient thing either expects some kind of different behavior from setting SO_BSDCOMPAT non-zero, or really doesn't rely on anything at all.
Since SO_BSDCOMPAT has had zero side effects for 5 or so years, this means that the safe change is to remove all references to SO_BSDCOMPAT that exist in any application.
So, the SO_BSDCOMPAT messages are in all likelihood unrelated to the problems I'm seeing with bind-9.2.1 under 2.5.6x-recent kernels...
I guess I'll have to turn up the debugging on bind and see if anything unusual pops up -
Joe
Quoted 40 lines · David S. Miller
Quoted 7 lines · Andrew Morton
The changelog has:
# --------------------------------------------
# 03/03/08 [email protected] 1.1083
# [NET]: NukeSO_BSDCOMPAT.
# --------------------------------------------Quoted 6 lines
Maybe James can tell us what is going on here.
We should at least place a cap on the number of times that message is printed.
Feel free to send a patch for that.
SO_BSDCOMPAThas had ZERO side effects since 2.0.x, and it's been thus scheduled to be removed for years. It was merely a binary state passed in and out of the kernel to the user and had no effect on socket behavior at all.Any application still referencing this ancient thing either expects some kind of different behavior from setting
SO_BSDCOMPATnon-zero, or really doesn't rely on anything at all.Since
SO_BSDCOMPAThas had zero side effects for 5 or so years, this means that the safe change is to remove all references toSO_BSDCOMPATthat exist in any application.-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: jjs <[email protected]>
Date: Wed, 12 Mar 2003 14:28:02 -0800
So, the SO_BSDCOMPAT messages are in all likelihood unrelated to the problems I'm seeing with bind-9.2.1 under 2.5.6x-recent kernels...
I guess I'll have to turn up the debugging on
bind and see if anything unusual pops up -
If bind errors internally because it cannot
set SO_BSDCOMPAT, this is likely the problem.
You need to hack the bind sources to remove references to SO_BSDCOMPAT.
Quoted 6 lines · David S. Miller
If bind errors internally because it cannot
setSO_BSDCOMPAT, this is likely the problem.You need to hack the bind sources to remove references to
SO_BSDCOMPAT.
Aha (light bulb goes on) -
OK, I'll take a look at the bind code and see...
Thanks for the clue -
Joe