“Bad Read-Header-Line header: #<eof>” while substituting

DoneSubmitted by Ludovic Courtès.
Details
3 participants
  • Ludovic Courtès
  • Ludovic Courtès
  • Christopher Baines
Owner
unassigned
Severity
serious
L
L
Ludovic Courtès wrote on 15 Mar 2021 15:26
“Bad Read-Header-Line header: #<e of>” while substituting
(address . bug-guix@gnu.org)
87eeggh4rh.fsf@inria.fr
As reported by a few people on IRC, ‘guix substitute’ sometimes fails in
a way that I just experienced (from
8154beffd8c121e953a7c4cd75c3eebfcc073a9a):

Toggle snippet (46 lines)
downloading from https://ci.guix.gnu.org/nar/gzip/0bji0q5n59595xaqkqrp2gv52lbz55xz-libpng-1.6.37 .
libpng-1.6.37 275KiB 11.0MiB/s 00:00 [##################] 100.0%

downloading from https://ci.guix.gnu.org/nar/lzip/h3a5ygxxh4gakhnl53mq7z9b43l8z05g-python-minimal.
python-minimal-wrapper-3.8.2 351B 293KiB/s 00:00 [##################] 100.0%

downloading from https://ci.guix.gnu.org/nar/lzip/h8j09yb5d8dh3jffvpzawxslig9bwhdr-freetype-2.10..
freetype-2.10.4 600KiB 3.0MiB/s 00:00 [##################] 100.0%

building /gnu/store/2wfzazqz9g5xizi4vq4pv75nkh1m24bp-perl-5.30.2.drv...
Backtrace:
In guix/ui.scm:
2164:12 19 (run-guix-command _ . _)
In guix/scripts/substitute.scm:
691:2 18 (guix-substitute . _)
In unknown file:
17 (with-continuation-barrier #<procedure thunk ()>)
In ice-9/boot-9.scm:
1736:10 16 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
In unknown file:
15 (apply-smob/0 #<thunk 7fcb3b3594a0>)
In ice-9/boot-9.scm:
1736:10 14 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
1736:10 13 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
1731:15 12 (with-exception-handler #<procedure 7fcb3a7c4150 at ice-9/boot-9.scm:1815:7 (exn)> _)
In guix/scripts/substitute.scm:
740:17 11 (_)
434:7 10 (process-substitution _ "/gnu/store/ns00dyapjbq9037dwrxa7hc31dvir00n-grub-minimal-2.)
In ice-9/boot-9.scm:
1736:10 9 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
In guix/scripts/substitute.scm:
443:9 8 (_)
In ice-9/boot-9.scm:
1731:15 7 (with-exception-handler #<procedure 7fcb3a7d37b0 at ice-9/boot-9.scm:1815:7 (exn)> _)
1669:16 6 (raise-exception _ #:continuable? _)
1667:16 5 (raise-exception _ #:continuable? _)
1669:16 4 (raise-exception _ #:continuable? _)
1764:13 3 (_ #<&compound-exception components: (#<&error> #<&irritants irritants: (read-header)
1669:16 2 (raise-exception _ #:continuable? _)
1667:16 1 (raise-exception _ #:continuable? _)
1669:16 0 (raise-exception _ #:continuable? _)

ice-9/boot-9.scm:1669:16: In procedure raise-exception:
Bad Read-Header-Line header: #<eof>

This is the kind of issue that ‘with-cached-connection’ as it can be
seen in 9158020d7853b6e7925802e0d0a082801c680e8f avoided:

Toggle snippet (23 lines)
(define* (call-with-cached-connection uri proc
#:optional
(open-connection
open-connection-for-uri/cached))
(let ((port (open-connection uri)))
(catch #t
(lambda ()
(proc port))
(lambda (key . args)
;; If PORT was cached and the server closed the connection in the
;; meantime, we get EPIPE. In that case, open a fresh connection and
;; retry. We might also get 'bad-response or a similar exception from
;; (web response) later on, once we've sent the request, or a
;; ERROR/INVALID-SESSION from GnuTLS.
(if (or (and (eq? key 'system-error)
(= EPIPE (system-error-errno `(,key ,@args))))
(and (eq? key 'gnutls-error)
(eq? (first args) error/invalid-session))
(memq key '(bad-response bad-header bad-header-component)))
(proc (open-connection uri #:fresh? #t))
(apply throw key args))))))

I think 7b812f7c84c43455cdd68a0e51b6ded018afcc8e and subsequent commits
may have caused this regression. In particular, in
20c08a8a45d0f137ead7c05e720456b2aea44402,
‘call-with-connection-error-handling’ is now used, but that one doesn’t
catch the exceptions mentioned above, in this case ‘bad-header’.

Ludo’.
L
L
Ludovic Courtès wrote on 15 Mar 2021 15:31
control message for bug #47157
(address . control@debbugs.gnu.org)
87czw0h4j3.fsf@gnu.org
severity 47157 serious
quit
C
C
Christopher Baines wrote on 15 Mar 2021 18:02
Re: bug#47157: “Bad Read-Header-Line header: #<eof>” while substituting
(name . Ludovic Courtès)(address . ludovic.courtes@inria.fr)(address . 47157@debbugs.gnu.org)
871rcgfiz9.fsf@cbaines.net
Ludovic Courtès <ludovic.courtes@inria.fr> writes:

Toggle quote (84 lines)
> As reported by a few people on IRC, ‘guix substitute’ sometimes fails in
> a way that I just experienced (from
> 8154beffd8c121e953a7c4cd75c3eebfcc073a9a):
>
> --8<---------------cut here---------------start------------->8---
> downloading from https://ci.guix.gnu.org/nar/gzip/0bji0q5n59595xaqkqrp2gv52lbz55xz-libpng-1.6.37 .
> libpng-1.6.37 275KiB 11.0MiB/s 00:00 [##################] 100.0%
>
> downloading from https://ci.guix.gnu.org/nar/lzip/h3a5ygxxh4gakhnl53mq7z9b43l8z05g-python-minimal.
> python-minimal-wrapper-3.8.2 351B 293KiB/s 00:00 [##################] 100.0%
>
> downloading from https://ci.guix.gnu.org/nar/lzip/h8j09yb5d8dh3jffvpzawxslig9bwhdr-freetype-2.10..
> freetype-2.10.4 600KiB 3.0MiB/s 00:00 [##################] 100.0%
>
> building /gnu/store/2wfzazqz9g5xizi4vq4pv75nkh1m24bp-perl-5.30.2.drv...
> Backtrace:
> In guix/ui.scm:
> 2164:12 19 (run-guix-command _ . _)
> In guix/scripts/substitute.scm:
> 691:2 18 (guix-substitute . _)
> In unknown file:
> 17 (with-continuation-barrier #<procedure thunk ()>)
> In ice-9/boot-9.scm:
> 1736:10 16 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
> In unknown file:
> 15 (apply-smob/0 #<thunk 7fcb3b3594a0>)
> In ice-9/boot-9.scm:
> 1736:10 14 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
> 1736:10 13 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
> 1731:15 12 (with-exception-handler #<procedure 7fcb3a7c4150 at ice-9/boot-9.scm:1815:7 (exn)> _)
> In guix/scripts/substitute.scm:
> 740:17 11 (_)
> 434:7 10 (process-substitution _ "/gnu/store/ns00dyapjbq9037dwrxa7hc31dvir00n-grub-minimal-2.)
> In ice-9/boot-9.scm:
> 1736:10 9 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
> In guix/scripts/substitute.scm:
> 443:9 8 (_)
> In ice-9/boot-9.scm:
> 1731:15 7 (with-exception-handler #<procedure 7fcb3a7d37b0 at ice-9/boot-9.scm:1815:7 (exn)> _)
> 1669:16 6 (raise-exception _ #:continuable? _)
> 1667:16 5 (raise-exception _ #:continuable? _)
> 1669:16 4 (raise-exception _ #:continuable? _)
> 1764:13 3 (_ #<&compound-exception components: (#<&error> #<&irritants irritants: (read-header)
> 1669:16 2 (raise-exception _ #:continuable? _)
> 1667:16 1 (raise-exception _ #:continuable? _)
> 1669:16 0 (raise-exception _ #:continuable? _)
>
> ice-9/boot-9.scm:1669:16: In procedure raise-exception:
> Bad Read-Header-Line header: #<eof>
> --8<---------------cut here---------------end--------------->8---
>
> This is the kind of issue that ‘with-cached-connection’ as it can be
> seen in 9158020d7853b6e7925802e0d0a082801c680e8f avoided:
>
> --8<---------------cut here---------------start------------->8---
> (define* (call-with-cached-connection uri proc
> #:optional
> (open-connection
> open-connection-for-uri/cached))
> (let ((port (open-connection uri)))
> (catch #t
> (lambda ()
> (proc port))
> (lambda (key . args)
> ;; If PORT was cached and the server closed the connection in the
> ;; meantime, we get EPIPE. In that case, open a fresh connection and
> ;; retry. We might also get 'bad-response or a similar exception from
> ;; (web response) later on, once we've sent the request, or a
> ;; ERROR/INVALID-SESSION from GnuTLS.
> (if (or (and (eq? key 'system-error)
> (= EPIPE (system-error-errno `(,key ,@args))))
> (and (eq? key 'gnutls-error)
> (eq? (first args) error/invalid-session))
> (memq key '(bad-response bad-header bad-header-component)))
> (proc (open-connection uri #:fresh? #t))
> (apply throw key args))))))
> --8<---------------cut here---------------end--------------->8---
>
> I think 7b812f7c84c43455cdd68a0e51b6ded018afcc8e and subsequent commits
> may have caused this regression. In particular, in
> 20c08a8a45d0f137ead7c05e720456b2aea44402,
> ‘call-with-connection-error-handling’ is now used, but that one doesn’t
> catch the exceptions mentioned above, in this case ‘bad-header’.

I think the behaviour changed unintentionally with [1], however,
thinking about the connection reuse in process-substitution compared
with http-multiple-get, there's no attempt here to look at if the server
has specified whether the connection should be closed.


Just like http-multiple-get, it's probably worth trying to check the
headers of the response, look at whether the server has indicated that
the connection should be closed, and if so, close the connection,
forcing a new one to be established for future requests.

I haven't tested this theory, but maybe if that happened, then some
occurrences of trying to read a response, and not being able to would be
prevented.
-----BEGIN PGP SIGNATURE-----
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=w5/u
-----END PGP SIGNATURE-----

C
C
Christopher Baines wrote on 15 Mar 2021 20:53
(address . 47157@debbugs.gnu.org)(name . Ludovic Courtès)(address . ludovic.courtes@inria.fr)
87v99sdwh9.fsf@cbaines.net
Christopher Baines <mail@cbaines.net> writes:

Toggle quote (102 lines)
> Ludovic Courtès <ludovic.courtes@inria.fr> writes:
>
>> As reported by a few people on IRC, ‘guix substitute’ sometimes fails in
>> a way that I just experienced (from
>> 8154beffd8c121e953a7c4cd75c3eebfcc073a9a):
>>
>> --8<---------------cut here---------------start------------->8---
>> downloading from https://ci.guix.gnu.org/nar/gzip/0bji0q5n59595xaqkqrp2gv52lbz55xz-libpng-1.6.37 .
>> libpng-1.6.37 275KiB 11.0MiB/s 00:00 [##################] 100.0%
>>
>> downloading from https://ci.guix.gnu.org/nar/lzip/h3a5ygxxh4gakhnl53mq7z9b43l8z05g-python-minimal.
>> python-minimal-wrapper-3.8.2 351B 293KiB/s 00:00 [##################] 100.0%
>>
>> downloading from https://ci.guix.gnu.org/nar/lzip/h8j09yb5d8dh3jffvpzawxslig9bwhdr-freetype-2.10..
>> freetype-2.10.4 600KiB 3.0MiB/s 00:00 [##################] 100.0%
>>
>> building /gnu/store/2wfzazqz9g5xizi4vq4pv75nkh1m24bp-perl-5.30.2.drv...
>> Backtrace:
>> In guix/ui.scm:
>> 2164:12 19 (run-guix-command _ . _)
>> In guix/scripts/substitute.scm:
>> 691:2 18 (guix-substitute . _)
>> In unknown file:
>> 17 (with-continuation-barrier #<procedure thunk ()>)
>> In ice-9/boot-9.scm:
>> 1736:10 16 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
>> In unknown file:
>> 15 (apply-smob/0 #<thunk 7fcb3b3594a0>)
>> In ice-9/boot-9.scm:
>> 1736:10 14 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
>> 1736:10 13 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
>> 1731:15 12 (with-exception-handler #<procedure 7fcb3a7c4150 at ice-9/boot-9.scm:1815:7 (exn)> _)
>> In guix/scripts/substitute.scm:
>> 740:17 11 (_)
>> 434:7 10 (process-substitution _ "/gnu/store/ns00dyapjbq9037dwrxa7hc31dvir00n-grub-minimal-2.)
>> In ice-9/boot-9.scm:
>> 1736:10 9 (with-exception-handler _ _ #:unwind? _ #:unwind-for-type _)
>> In guix/scripts/substitute.scm:
>> 443:9 8 (_)
>> In ice-9/boot-9.scm:
>> 1731:15 7 (with-exception-handler #<procedure 7fcb3a7d37b0 at ice-9/boot-9.scm:1815:7 (exn)> _)
>> 1669:16 6 (raise-exception _ #:continuable? _)
>> 1667:16 5 (raise-exception _ #:continuable? _)
>> 1669:16 4 (raise-exception _ #:continuable? _)
>> 1764:13 3 (_ #<&compound-exception components: (#<&error> #<&irritants irritants: (read-header)
>> 1669:16 2 (raise-exception _ #:continuable? _)
>> 1667:16 1 (raise-exception _ #:continuable? _)
>> 1669:16 0 (raise-exception _ #:continuable? _)
>>
>> ice-9/boot-9.scm:1669:16: In procedure raise-exception:
>> Bad Read-Header-Line header: #<eof>
>> --8<---------------cut here---------------end--------------->8---
>>
>> This is the kind of issue that ‘with-cached-connection’ as it can be
>> seen in 9158020d7853b6e7925802e0d0a082801c680e8f avoided:
>>
>> --8<---------------cut here---------------start------------->8---
>> (define* (call-with-cached-connection uri proc
>> #:optional
>> (open-connection
>> open-connection-for-uri/cached))
>> (let ((port (open-connection uri)))
>> (catch #t
>> (lambda ()
>> (proc port))
>> (lambda (key . args)
>> ;; If PORT was cached and the server closed the connection in the
>> ;; meantime, we get EPIPE. In that case, open a fresh connection and
>> ;; retry. We might also get 'bad-response or a similar exception from
>> ;; (web response) later on, once we've sent the request, or a
>> ;; ERROR/INVALID-SESSION from GnuTLS.
>> (if (or (and (eq? key 'system-error)
>> (= EPIPE (system-error-errno `(,key ,@args))))
>> (and (eq? key 'gnutls-error)
>> (eq? (first args) error/invalid-session))
>> (memq key '(bad-response bad-header bad-header-component)))
>> (proc (open-connection uri #:fresh? #t))
>> (apply throw key args))))))
>> --8<---------------cut here---------------end--------------->8---
>>
>> I think 7b812f7c84c43455cdd68a0e51b6ded018afcc8e and subsequent commits
>> may have caused this regression. In particular, in
>> 20c08a8a45d0f137ead7c05e720456b2aea44402,
>> ‘call-with-connection-error-handling’ is now used, but that one doesn’t
>> catch the exceptions mentioned above, in this case ‘bad-header’.
>
> I think the behaviour changed unintentionally with [1], however,
> thinking about the connection reuse in process-substitution compared
> with http-multiple-get, there's no attempt here to look at if the server
> has specified whether the connection should be closed.
>
> 1: https://git.savannah.gnu.org/cgit/guix.git/commit/?id=f50f5751fff4cfc6d5abba9681054569694b7a5c
>
> Just like http-multiple-get, it's probably worth trying to check the
> headers of the response, look at whether the server has indicated that
> the connection should be closed, and if so, close the connection,
> forcing a new one to be established for future requests.
>
> I haven't tested this theory, but maybe if that happened, then some
> occurrences of trying to read a response, and not being able to would be
> prevented.

I've now actually got around to testing this, I'm no expert at running
the substitute script manually without the guix-daemon, but I gave it a
go, using a local NGinx instance which just allowed two requests per
connection.

With these changes [2], connections were closed when appropriate and a
new one is established when the next nar is fetched.


When testing using the same approach with master, I get this exception
[3], so either I'm doing something wrong, or this isn't a circumstance
under which there can be a bad-header issue. Regardless, I'm not sure
what's going on with this exception below.

3:
ice-9/boot-9.scm:1669:16: In procedure raise-exception:
Wrong number of values returned to continuation (expected 2)
-----BEGIN PGP SIGNATURE-----
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=Fb25
-----END PGP SIGNATURE-----

L
L
Ludovic Courtès wrote on 15 Mar 2021 21:30
(name . Christopher Baines)(address . mail@cbaines.net)(address . 47157@debbugs.gnu.org)
87a6r4cg88.fsf@gnu.org
Christopher Baines <mail@cbaines.net> skribis:

Toggle quote (18 lines)
>> I think 7b812f7c84c43455cdd68a0e51b6ded018afcc8e and subsequent commits
>> may have caused this regression. In particular, in
>> 20c08a8a45d0f137ead7c05e720456b2aea44402,
>> ‘call-with-connection-error-handling’ is now used, but that one doesn’t
>> catch the exceptions mentioned above, in this case ‘bad-header’.
>
> I think the behaviour changed unintentionally with [1], however,
> thinking about the connection reuse in process-substitution compared
> with http-multiple-get, there's no attempt here to look at if the server
> has specified whether the connection should be closed.
>
> 1: https://git.savannah.gnu.org/cgit/guix.git/commit/?id=f50f5751fff4cfc6d5abba9681054569694b7a5c
>
> Just like http-multiple-get, it's probably worth trying to check the
> headers of the response, look at whether the server has indicated that
> the connection should be closed, and if so, close the connection,
> forcing a new one to be established for future requests.

I think that’s not enough because we can’t rely on the server’s state
intent here.

For example, you have a keep-alive connection that you keep in cache.
Minutes later, you come back and send a request over that port. If the
server dropped the connection in the meantime, that can manifest in any
of the ways we’ve seen: 'bad-response when attempting to read the
response, some 'gnutls-error, 'system-error and EPIPE, etc. There’s no
way to determine in advance whether the socket is fine.

That’s why the initial approach was to wrap all the call sites were the
socket was known to be possibly “tainted” in ‘with-cached-connection’.

Toggle quote (5 lines)
> I've now actually got around to testing this, I'm no expert at running
> the substitute script manually without the guix-daemon, but I gave it a
> go, using a local NGinx instance which just allowed two requests per
> connection.

I believe in this case ‘port-closed?’ returns true because the
socket/TLS record port got closed right at the end of the response, so
it’s the “easy” case; I don’t think it captures the situation I
described above where an error comes up later while trying to write
to/read from the port.

Ludo’.
C
C
Christopher Baines wrote on 15 Mar 2021 21:37
(name . Ludovic Courtès)(address . ludo@gnu.org)(address . 47157@debbugs.gnu.org)
87sg4wdugg.fsf@cbaines.net
Ludovic Courtès <ludo@gnu.org> writes:

Toggle quote (44 lines)
> Christopher Baines <mail@cbaines.net> skribis:
>
>>> I think 7b812f7c84c43455cdd68a0e51b6ded018afcc8e and subsequent commits
>>> may have caused this regression. In particular, in
>>> 20c08a8a45d0f137ead7c05e720456b2aea44402,
>>> ‘call-with-connection-error-handling’ is now used, but that one doesn’t
>>> catch the exceptions mentioned above, in this case ‘bad-header’.
>>
>> I think the behaviour changed unintentionally with [1], however,
>> thinking about the connection reuse in process-substitution compared
>> with http-multiple-get, there's no attempt here to look at if the server
>> has specified whether the connection should be closed.
>>
>> 1: https://git.savannah.gnu.org/cgit/guix.git/commit/?id=f50f5751fff4cfc6d5abba9681054569694b7a5c
>>
>> Just like http-multiple-get, it's probably worth trying to check the
>> headers of the response, look at whether the server has indicated that
>> the connection should be closed, and if so, close the connection,
>> forcing a new one to be established for future requests.
>
> I think that’s not enough because we can’t rely on the server’s state
> intent here.
>
> For example, you have a keep-alive connection that you keep in cache.
> Minutes later, you come back and send a request over that port. If the
> server dropped the connection in the meantime, that can manifest in any
> of the ways we’ve seen: 'bad-response when attempting to read the
> response, some 'gnutls-error, 'system-error and EPIPE, etc. There’s no
> way to determine in advance whether the socket is fine.
>
> That’s why the initial approach was to wrap all the call sites were the
> socket was known to be possibly “tainted” in ‘with-cached-connection’.
>
>> I've now actually got around to testing this, I'm no expert at running
>> the substitute script manually without the guix-daemon, but I gave it a
>> go, using a local NGinx instance which just allowed two requests per
>> connection.
>
> I believe in this case ‘port-closed?’ returns true because the
> socket/TLS record port got closed right at the end of the response, so
> it’s the “easy” case; I don’t think it captures the situation I
> described above where an error comes up later while trying to write
> to/read from the port.

Yeah, of course, I think error handling is needed as well, it just
occurred to me when looking at this issue and the relevant code.
-----BEGIN PGP SIGNATURE-----
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=A6Tc
-----END PGP SIGNATURE-----

L
L
Ludovic Courtès wrote on 18 Mar 2021 17:04
(name . Christopher Baines)(address . mail@cbaines.net)(address . 47157-done@debbugs.gnu.org)
8735wsmos2.fsf@gnu.org
This should now be fixed by c37e3b92ad0334ba2fe7ee4e98631f0a4edeee21.

Thanks, Chris!

Ludo’.
Closed
?
Your comment

This issue is archived.

To comment on this conversation send email to 47157@debbugs.gnu.org