QA status: 660 is only failing test I have; I've made some
notes in commit log below and added extra diagnostics into
the test. Its a new pmwebd test, so not a regression, and
hopefully will get resolved soon.
Not run: 009 066 069 081 083 086 117 122 130 148 159 164 212 215 223 233 262
272 307 308 311 353 360 361 365 369 370 371 372 374 375 377 420 446 476 477 478
480 504 512 519 545 570 583 640 652 654 704 717 718
Code status: there were a couple of conflicts to sort out,
please review (esp. kenj and brolley re net wildcards).
Changes committed to git://oss.sgi.com/pcp/pcp.git dev
man/man1/pcpintro.1 | 9
qa/.gitignore | 2
qa/023 | 26
qa/066.out.4 | 4
qa/069 | 12
qa/069.out.5 | 73 ++
qa/069.out.6 | 71 ++
qa/117 | 2
qa/200 | 2
qa/200.out | 46 -
qa/200.out.2 | 46 -
qa/244 | 32 -
qa/251 | 4
qa/287 | 12
qa/367.out.2 | 9
qa/403 | 4
qa/417 | 4
qa/433 | 4
qa/457 | 32 -
qa/532 | 3
qa/578 | 4
qa/589 | 28
qa/660 | 13
qa/660.out | 58 -
qa/660.py | 96 ---
qa/712 | 4
qa/713 | 6
qa/714 | 4
qa/715 | 28
qa/715.out | 7
qa/716 | 8
qa/716.out | 86 --
qa/716.out.1 | 86 ++
qa/716.out.2 | 88 ++
qa/717 | 2
qa/718 | 2
qa/common.check | 9
qa/common.secure | 9
qa/src/GNUlocaldefs | 2
qa/src/test_webapi.python | 114 +++
src/libpcp/src/GNUmakefile | 1
src/libpcp/src/access.c | 1275 ++++++++++++++++++++++++++++---------------
src/libpcp/src/auxserver.c | 10
src/libpcp/src/check-statics | 3
src/libpcp/src/nss_connect.c | 16
src/libpcp/src/nss_server.c | 12
src/pmwebapi/pmresapi.c | 19
src/pmwebapi/pmwebapi.c | 2
48 files changed, 1522 insertions(+), 867 deletions(-)
commit cfb431865b36313d2394e2499402312588f7a896
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 20:18:47 2013 +1000
Filter both pcp stop and start for any pcp restart in qa
commit bb458fdb841acc9954d03c7b39d6634608f643a3
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 20:18:00 2013 +1000
Missed addition of -e option to Install on prior commits of qa/715
commit 8b368707fa815f14164ab0001a4c4ee2a4ad3ed9
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 20:17:34 2013 +1000
Correct the check for verbosity in check_agent
commit b484af058ba0af37a0042b49574cea83295f5a65
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 18:02:18 2013 +1000
Update qa/715 to use the common pmcd setup switch/restore code
commit f004063b1118fd96cca9014a11940d406d195a3e
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 16:33:40 2013 +1000
Additional diagnostics in the webapi python test script
Seeing test failure in webd test 660:
660 - output mismatch (see 660.out.bad)
20c20
< test #### metric .* enumeration match count PASS
---
> test #### metric .* enumeration match count FAIL
41c41
< test #### metric .* enumeration match count PASS
---
> test #### metric .* enumeration match count FAIL
Added some diagnostics to understand the comparisons,
now seeing the two failure cases: local-context mode
(for pminfo-based query, I also get proc.* and pmcd.*
metrics, which are not there for the pmwebd query),
and then in host-context mode, it seems pmwebd is not
handling sample.bad.unknown (missing in the output).
commit 81bedcee0d1700c27c64a886e2e8a4783970f5ad
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 15:29:24 2013 +1000
Update qa/660 to check 3800 pcp and cull its .out
commit 01a20d757f2e885d4a1696e425ebc35d40bcd9e3
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 15:25:10 2013 +1000
Handle different pmdasimple indom text output, after code fix
commit 8c18bacf7c97fccfa8f8aa7dde29c5a34a456b3b
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 15:19:09 2013 +1000
Filter optional pmlogger restart messages from secure-sockets tests
commit ab465de5580e8b5e25265ab87e4666fd47449166
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 14:29:30 2013 +1000
Add notrun checks to tests 717 and 718 for pmdasimple python variants
commit 167ed712e2eb593791d63c8ce0ac18a0e56f9be9
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 14:27:01 2013 +1000
Resolve fallout from sharing check_agent routine in qa/578
commit 63b320622303c27e98e46e6f7569d2a4b753fc91
Merge: 604c793 f28fb13
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 11:20:29 2013 +1000
Merge branch 'fche/pmwebapi' of ../pcpfans into dev
commit 604c793368e1fab413bb82cc9bdbc03f85e8641c
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 11:19:19 2013 +1000
Rename python test script to match others and ensure build passiness
The py extension is magic to certain package managers, and if used in
a script name causes follow-on effects (such as byte-compilation) and
at the moment, we don't cater for this.
error: Installed (but unpackaged) file(s) found:
/var/lib/pcp/testsuite/660.pyc
/var/lib/pcp/testsuite/660.pyo
For now, use the same approach as the other python scripts to avoid
this pitfall.
commit f28fb1302a0057391d654624c62a1ef1997d0cca
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date: Sun Apr 28 21:16:17 2013 -0400
pmwebapi: catch pmresapi file ../FOO as invalid
The previous version looked for /../ only in the URL, not in the
concatenated BASEDIR+/+URL string.
commit 49dc42f982757f82fbadbd124ddff2751128e23e
Merge: a1b7b1e 16034a0
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 11:04:35 2013 +1000
Merge branch 'fche/pmwebapi' of ../pcpfans into dev
commit 16034a0f057ba4ff6f688cdf76b8321cb009cd93
Author: Frank Ch. Eigler <fche@xxxxxxxxxx>
Date: Sun Apr 28 21:01:29 2013 -0400
pmwebapi: error-response thinko fix
A previous version mistook the mime content-type header as something
associated with a connection (even though specified by a resp*), and
in one case erroneously dereferenced an uninitialized pointer.
commit a1b7b1e3984f6889a83b42ae218615cf2d389249
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 10:42:11 2013 +1000
Add temporary LDIRT target for old nss-named sources
commit 0db071b4650c7eee6f2b9cd4b536bab0babd5951
Merge: 4cbf740 997c51f
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 10:39:59 2013 +1000
Merge branch 'fche/pmwebapi' of ../pcpfans into dev
commit 4cbf740d2f16b26af7bb4080bce26784bcc6f4cb
Merge: 8a33e7f 821ddb4
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 10:38:45 2013 +1000
Merge branch 'dev' of ../nathans-pcp into dev
Conflicts:
src/libpcp/src/auxserver.c
(Resolved by taking nathans version of with-ipv6 handling)
commit 8a33e7fc78f96ecf3b661288b421e08e174d37ec
Merge: 2b45ca4 1767dc7
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 10:31:01 2013 +1000
Merge branch 'brolley/dev' of ../pcpfans into dev
Conflicts:
src/libpcp/src/access.c
(Conflict resolved via email advice from Dave)
commit 821ddb4e645b60a693fccabad0a8afbbe97ef3f5
Author: Nathan Scott <nathans@xxxxxxxxxx>
Date: Mon Apr 29 08:53:52 2013 +1000
Add correct 367 output with userauth PDUs
commit 1767dc7a35ecd79bbe90cd28cef17f47831bb658
Merge: fe59d3b 88a5619
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Apr 25 15:05:22 2013 -0400
Merge branch 'brolley/dev' of ssh://sourceware.org/git/pcpfans into
brolley/dev
Conflicts:
qa/200
qa/200.out
qa/200.out.2
commit fe59d3b8ea11131657ce7bc4ce2266a0703b508d
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Apr 25 11:50:23 2013 -0400
Prevent buffer overflow when parsing access wildcards.
commit 0817d426dcb9797f3ab98ffd0a1da9c93b54813c
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Apr 25 11:21:00 2013 -0400
Incorporate the following change from Ken McDonell:
libpcp/__pmAccAddHost - don't always create ipv6 wildcard entry
If the platform does not support IPv6, don't create IPv6 wildcard
entries alongside IPv4 entries.
commit cec10e970d2de2421fbf7505da5945bf5d5b7c46
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Wed Apr 24 17:13:35 2013 -0400
__pmAccAddHost(), __pmAccAddClient() and __pmAccDelClient() were
only considering one network address for a given host. However
a given host can have an arbitrary number of addresses due to
multiple network interfaces and multiple address families
configured on each (e.g. inet, ipv6).
This is a potential security issue since, when a host is specified
by name, the user presumably wishes to restrict all access from that
host but, currently, only one address associated with that host is
being restricted.
In this commit:
- __pmAccAddHost() now adds all of the network addresses associated
with a given host, specified by name, to the host access table.
Hosts specified directly by address continue to generate a single
entry in the table.
- __pmAccAddClient accumulates the permissions for all network
addresses associated with a client specified as "localhost".
- __pmDelClient deletes entries for all network addresses associated
with a client specified as "localhost". It also now maps "localhost"
to the actual host as __pmAccAddHost() and __pmAccAddClient() already
did (and still do).
- qa tests 023 and 244: filter the host access table to include
only one entry for localhost.
commit 88a56192add7d3916907bf3ef1794ecfa2251659
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Apr 25 11:50:23 2013 -0400
Prevent buffer overflow when parsing access wildcards.
commit 881be9e238a7e509a70f54090f67e05855477291
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Thu Apr 25 11:21:00 2013 -0400
Incorporate the following change from Ken McDonell:
libpcp/__pmAccAddHost - don't always create ipv6 wildcard entry
If the platform does not support IPv6, don't create IPv6 wildcard
entries alongside IPv4 entries.
commit 2b45ca49944d36125ef4f47f7620bd80f027418f
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Apr 25 14:22:14 2013 +1000
qa/532 - add note about non-deterministic failures
commit 980212a94bf8440d8e47bff2dca4af3764dc22df
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Apr 25 14:18:12 2013 +1000
qa/589 - explain port 26, handle firewall rejection
Different error message from some firewalls.
commit 6fd2465452f5a4be8abc1f2941cb6f3ed80f9eeb
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Apr 25 13:51:37 2013 +1000
qa/common.check - remove unwanted diagnostic tmp file save
commit bb595464a5e2221dc4bb1bbdedfae9a699ed337a
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Apr 25 13:50:48 2013 +1000
qa/069 - new output for post PCP 3.7.2 and no IPv6 support
commit 3a272a2020f4cea1544f9538f1b3647eeb742ec5
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Apr 25 12:19:58 2013 +1000
libpcp - use our wrapper for run-time check on IPv6 support
Since __pmGetAPIConfig("ipv6") does the job, no need to expose
the ugly platform dependent code elsewhere.
commit eba78baa2c1ff6e2a35e77dfe37d54dda2d9096a
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Thu Apr 25 12:12:43 2013 +1000
libpcp/__pmAccAddHost - don't always create ipv6 wildcard entry
If the platform does not support IPv6, don't create IPv6 wildcard
entries alongside IPv4 entries.
Fix related qa/066.out.4 output (for platforms that don't support
IPv6) as a result of an eariler for minor formatting change in
reporting IPv4 wildcards.
commit 99716f0a87a4ea8847139ed4a44c627bec655a31
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Wed Apr 24 17:13:35 2013 -0400
__pmAccAddHost(), __pmAccAddClient() and __pmAccDelClient() were
only considering one network address for a given host. However
a given host can have an arbitrary number of addresses due to
multiple network interfaces and multiple address families
configured on each (e.g. inet, ipv6).
This is a potential security issue since, when a host is specified
by name, the user presumably wishes to restrict all access from that
host but, currently, only one address associated with that host is
being restricted.
In this commit:
- __pmAccAddHost() now adds all of the network addresses associated
with a given host, specified by name, to the host access table.
Hosts specified directly by address continue to generate a single
entry in the table.
- __pmAccAddClient accumulates the permissions for all network
addresses associated with a client specified as "localhost".
- __pmDelClient deletes entries for all network addresses associated
with a client specified as "localhost". It also now maps "localhost"
to the actual host as __pmAccAddHost() and __pmAccAddClient() already
did (and still do).
- qa tests 023 and 244: filter the host access table to include
only one entry for localhost.
commit 77b6b114bb7d86ad6366452fe59e9c91c0f9c10b
Author: Dave Brolley <brolley@xxxxxxxxxx>
Date: Wed Apr 24 17:12:35 2013 -0400
QA test 200: Filter 127.0.0.1 and ::1 to LOOPBACK.
commit 2d6f43fd2c0a707d876323f1af92c4c8764cbd70
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Apr 23 20:16:28 2013 +1000
qa/287 - fix up last botched commits
Now have Mac OS X _and_ uname changes.
commit 574fbb411cf87415722aaa9774e6930159040b27
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Apr 23 20:06:19 2013 +1000
Revert "qa/287 - uname -m more reliable than uname -p"
This reverts commit 0b50ff773f613e8194af7e25066b9f84508d8204.
Actually this commit accidently undid the previous commit ...
so put that back, and then we'll redo the uname change.
commit 0af1ef8cfeeffd13c2be05e14ff9050ee5ee6c22
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Apr 23 19:52:26 2013 +1000
qa assorted changes to expunge echo -n
The following is NOT portable
$ echo -n 'foo bar'
we introduced support for doing this in a way that works a long
time ago, so rewrite the above to be
$PCP_ECHO_PROG $PCP_ECHO_N 'foo bar'"$PCP_ECHO_C"
Ugly I know, but works in the BSD-based and Unix-based worlds,
as well as Linux/GNU land.
commit ef41da6eaede837775f96a41082fe457b222a313
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Apr 23 16:30:45 2013 +1000
Introduce $PCP_SECURE_DB_METHOD environment var
This is an attempt to overcome the problems associated with
hardcoding the sql: method into the certificate and key database
management, which fails badly on older platforms, e.g. CentOS 5.9
(and probably other, non-Linux platforms) where the sql: method
is not supported and using export PCP_SECURE_DB_METHOD='' seems
to restore old functionality for the non-secure socket users.
Note that $NSS_DEFAULT_DB_TYPE is not an option as this apparently
intended for compatibility support in the other direction,
as "sql" appears to be the only well-defined value for this
environment variable to enable the "new" method.
commit c3f950b20ca4f6d7c7224136aef3c0090052c0e5
Merge: 12bccf4 b4106c3
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Tue Apr 23 16:29:43 2013 +1000
Merge branch 'dev' of git://oss.sgi.com/pcp/pcp into dev
commit 12bccf4f017cbe5bede74311b8de3319b847abae
Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Mon Apr 22 17:32:31 2013 +1000
qa/069 - learn about changes in ip addr wildcard reporting
Recent ipv6 changes + ipv4 format seems to have been changed for
ipv6 reporting consistency.
|