ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
LS
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
The integration tests don't work for me locally with these patches. Currently I have applied: $ git log --oneline origin/master..HEAD 782c839 SYSDB: Sanitize dn in sysdb_get_user_members_recursively a176315 SDAP: sysdb_search_users does not set users_count for failures 02e74e6 SDAP: sanitize member name before using in filter a40f2bd test_ldap: test resolving of names with special characters 300a6de test_ldap: test nested membership with rfc2307bis cf78e01 intg: Fix pep8 warnings 635c311 sssd_id.py: Primary group should be returned for initgroups e11c615 intg: create ldap test without enumeration eab68a7 test_enumeration: Remove test without enumeration 83b07af intg: rename test with enumeration
The Commit hashes are: $ git rev-list origin/master..HEAD 782c8394ae40bc5aba03c657f869fcba8ddb4c2d a1763150dce0ddc1ab9260f2f8447ec525976d43 02e74e6c4c33c023cb7a8608f80afd5a1c24a055 a40f2bd86001bd3696184850a84f59592d6362c9 300a6de00d4e3f71bb9afb9670495dfef43942fa cf78e01936e3c0c4a0cb3b4596215bfbb938a181 635c311d287d71fad1fe9eb0247385506a5ba781 e11c615f897f793275f4ca9870a7cdc45d14ff88 eab68a720a042572a4848aa5d5815e21e54a0afe 83b07af8602a6bea9a74acfefa04cfa0d0d8ae87
On (04/08/16 12:16), Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
The integration tests don't work for me locally with these patches. Currently I have applied:
May I know which tests are failing?
LS
On Thu, Aug 04, 2016 at 12:55:02PM +0200, Lukas Slebodnik wrote:
On (04/08/16 12:16), Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
The integration tests don't work for me locally with these patches. Currently I have applied:
May I know which tests are failing?
All of the memory cache tests were failing with "Cannot contact LDAP server" or similar. But strangely, everything works now..
I'm off to reviewing the patches in the meantime..
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I'll give a formal ACK once the CI run finishes.
On Tue, Aug 09, 2016 at 12:36:11PM +0200, Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I'll give a formal ACK once the CI run finishes.
btw while reading the patches I again realized the big amount of code duplication we have between tests. Most of the fixtures are just copy-pasted. This is not related to this patchset per-se, but I think that during the 1.14.2 development, we should focus on making the integration tests easier to run and debug. So we should have tickets to track the issues in tests and reducing code duplication should be one of them.
On Tue, Aug 09, 2016 at 12:36:11PM +0200, Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I'll give a formal ACK once the CI run finishes.
For some reason, I can't get a 'green' CI run on RHEL-6. I tried three times, the last attempt is here: http://sssd-ci.duckdns.org/logs/job/51/19/rhel6/ci-build-debug/ci-make-intgc... but all had the same errors.
Did the patches change something in the way the server is set up on RHEL-6?
On (09/08/16 14:54), Jakub Hrozek wrote:
On Tue, Aug 09, 2016 at 12:36:11PM +0200, Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I can remove it as part of "moving files" if you wish.
I'll give a formal ACK once the CI run finishes.
For some reason, I can't get a 'green' CI run on RHEL-6. I tried three times, the last attempt is here: http://sssd-ci.duckdns.org/logs/job/51/19/rhel6/ci-build-debug/ci-make-intgc... but all had the same errors.
Did the patches change something in the way the server is set up on RHEL-6?
E SERVER_DOWN: {'desc': "Can't contact LDAP server"}
I have no idea why it should be stopped. We do not stop openldap server. It should be done only in cleanup (end of all tests in module)
LS
On (09/08/16 19:12), Lukas Slebodnik wrote:
On (09/08/16 14:54), Jakub Hrozek wrote:
On Tue, Aug 09, 2016 at 12:36:11PM +0200, Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I can remove it as part of "moving files" if you wish.
I'll give a formal ACK once the CI run finishes.
For some reason, I can't get a 'green' CI run on RHEL-6. I tried three times, the last attempt is here: http://sssd-ci.duckdns.org/logs/job/51/19/rhel6/ci-build-debug/ci-make-intgc... but all had the same errors.
Did the patches change something in the way the server is set up on RHEL-6?
E SERVER_DOWN: {'desc': "Can't contact LDAP server"}
I have no idea why it should be stopped. We do not stop openldap server. It should be done only in cleanup (end of all tests in module)
It took me some time to find root of cause these failures. The 1st patch fixes the issues. And i got a recomendation from freeipa develoepers that it's better to avoid using destructors for cleanup. With pytest we should use finalizers or with keyword the same as for files.
LS
On Fri, Aug 12, 2016 at 05:12:44PM +0200, Lukas Slebodnik wrote:
On (09/08/16 19:12), Lukas Slebodnik wrote:
On (09/08/16 14:54), Jakub Hrozek wrote:
On Tue, Aug 09, 2016 at 12:36:11PM +0200, Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote:
ehlo,
attched patch is an integration test for regression #3093. I prepared a test and I let someone else to fix it :-)
I will try to find more bugs in downstream tests.
BTW we might move test from test_ts_cache somewhere else. But it was the fastest way how to write a test without enumeration.
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I can remove it as part of "moving files" if you wish.
I'll give a formal ACK once the CI run finishes.
For some reason, I can't get a 'green' CI run on RHEL-6. I tried three times, the last attempt is here: http://sssd-ci.duckdns.org/logs/job/51/19/rhel6/ci-build-debug/ci-make-intgc... but all had the same errors.
Did the patches change something in the way the server is set up on RHEL-6?
E SERVER_DOWN: {'desc': "Can't contact LDAP server"}
I have no idea why it should be stopped. We do not stop openldap server. It should be done only in cleanup (end of all tests in module)
It took me some time to find root of cause these failures. The 1st patch fixes the issues. And i got a recomendation from freeipa develoepers that it's better to avoid using destructors for cleanup. With pytest we should use finalizers or with keyword the same as for files.
LS
Thank you for fixing all this. CI passed for me both locally and in Jenkins: http://sssd-ci.duckdns.org/logs/job/52/00/summary.html
ACK to all
On (18/08/16 11:37), Jakub Hrozek wrote:
On Fri, Aug 12, 2016 at 05:12:44PM +0200, Lukas Slebodnik wrote:
On (09/08/16 19:12), Lukas Slebodnik wrote:
On (09/08/16 14:54), Jakub Hrozek wrote:
On Tue, Aug 09, 2016 at 12:36:11PM +0200, Jakub Hrozek wrote:
On Wed, Aug 03, 2016 at 09:56:40AM +0200, Lukas Slebodnik wrote:
On (13/07/16 17:48), Lukas Slebodnik wrote: >ehlo, > >attched patch is an integration test for regression #3093. >I prepared a test and I let someone else to fix it :-) > >I will try to find more bugs in downstream tests. > >BTW we might move test from test_ts_cache somewhere else. >But it was the fastest way how to write a test without enumeration. >
As I previously wrote test_ts_cache is not the best place for this test. Attached are patches which some changes in intg tests. There is also test for grups with special characters
LS
Thank you, the only comment I have is that we should remove ldap_auth_disable_tls_never_use_in_production option. We don't use it in tests (we don't test authentication there at all at the moment) and it's better to not advertise the option.
I can remove it as part of "moving files" if you wish.
I'll give a formal ACK once the CI run finishes.
For some reason, I can't get a 'green' CI run on RHEL-6. I tried three times, the last attempt is here: http://sssd-ci.duckdns.org/logs/job/51/19/rhel6/ci-build-debug/ci-make-intgc... but all had the same errors.
Did the patches change something in the way the server is set up on RHEL-6?
E SERVER_DOWN: {'desc': "Can't contact LDAP server"}
I have no idea why it should be stopped. We do not stop openldap server. It should be done only in cleanup (end of all tests in module)
It took me some time to find root of cause these failures. The 1st patch fixes the issues. And i got a recomendation from freeipa develoepers that it's better to avoid using destructors for cleanup. With pytest we should use finalizers or with keyword the same as for files.
LS
Thank you for fixing all this. CI passed for me both locally and in Jenkins: http://sssd-ci.duckdns.org/logs/job/52/00/summary.html
ACK to all
master: * e59b73366d3067c576e39a214a34ace2f9a84878 * 1ba26271952de1beeb9e041bedf87d720d3f5680 * a949dfb6b03c70896e6ab3c7a10781e8ecbaadc2 * de19c0af27f1576c13bef183600136851baf767d * 49a5412cbc98e630de17359c29cb8d6ce0e16168 * 8def4f07aedf2cebf77e816b201d91beac9b964b * 1f4dc2971bac4ceb0803b18f86a746656a0f1990 * 361f9fe207f82693bbfd40dfe272a4513652856c * a035366f622f42c865dbaa1579f741a81548fde2
LS
sssd-devel@lists.fedorahosted.org