Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote:
Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
Maybe we can do this during the init phase of the responder? We currently only call sysdb_update_subdomains() here. If we want to add sss_write_domain_mappings() and maybe adding ads_store_sdap_subdom() there are well would be equivalent to your patch?
bye, Sumit
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote:
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote:
Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
Maybe we can do this during the init phase of the responder? We currently only call sysdb_update_subdomains() here.
I think that's a good idea, I'll prepare a patch.
If we want to add sss_write_domain_mappings() and maybe adding ads_store_sdap_subdom() there are well would be equivalent to your patch?
Sorry, I don't quite understand this part, I don't think we can call ads_store_sdap_subdom() the provider doesn't know what kind of backend it talks to..
On Mon, Dec 16, 2013 at 07:03:11PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote:
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote:
Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
Maybe we can do this during the init phase of the responder? We currently only call sysdb_update_subdomains() here.
I think that's a good idea, I'll prepare a patch.
If we want to add sss_write_domain_mappings() and maybe adding ads_store_sdap_subdom() there are well would be equivalent to your patch?
Sorry, I don't quite understand this part, I don't think we can call ads_store_sdap_subdom() the provider doesn't know what kind of backend it
^^^^^^^^ I meant to say "responder" here.
On Mon, Dec 16, 2013 at 07:05:16PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 07:03:11PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote:
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote:
Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
Maybe we can do this during the init phase of the responder? We
ah, sorry, I had the typo here ^^^^^^^^ first. I meant to call it during the initialization of the subdomain part of the AD and IPA providers.
bye, Sumit
currently only call sysdb_update_subdomains() here.
I think that's a good idea, I'll prepare a patch.
If we want to add sss_write_domain_mappings() and maybe adding ads_store_sdap_subdom() there are well would be equivalent to your patch?
Sorry, I don't quite understand this part, I don't think we can call ads_store_sdap_subdom() the provider doesn't know what kind of backend it
^^^^^^^^ I meant to say "responder" here.
sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://lists.fedorahosted.org/mailman/listinfo/sssd-devel
On Mon, Dec 16, 2013 at 10:09:58PM +0100, Sumit Bose wrote:
On Mon, Dec 16, 2013 at 07:05:16PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 07:03:11PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote:
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote:
Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
Maybe we can do this during the init phase of the responder? We
ah, sorry, I had the typo here ^^^^^^^^ first. I meant to call it during the initialization of the subdomain part of the AD and IPA providers.
bye, Sumit
That makes perfect sense :-)
See the attached patch, I moved the code to a separate function.
On Tue, Dec 17, 2013 at 05:02:07PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 10:09:58PM +0100, Sumit Bose wrote:
On Mon, Dec 16, 2013 at 07:05:16PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 07:03:11PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote:
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote:
Hi,
I found this bug when testing the GC patches. Previously, when SSSD was started, but subdomains list was up-to-date, the ad_ctx was not initialized for the subdomain.
I was also thinking whether we should re-initialize the domain-realm mappings after sssd startup, the way we re-initialize kdcinfo files. I don't think it's strictly necessary because if someone deletes a file in /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we should be at least aware..
Maybe we can do this during the init phase of the responder? We
ah, sorry, I had the typo here ^^^^^^^^ first. I meant to call it during the initialization of the subdomain part of the AD and IPA providers.
bye, Sumit
That makes perfect sense :-)
See the attached patch, I moved the code to a separate function.
Actually, I did one more change in the original patch and also did a similar change in the IPA provider. The result is a second patch. I mostly split the patches into two because the IPA patch is a convenience, so from my point of view it's fine to fix it in master only. The first one is something I'd like in sssd-1-11, too.
On Tue, Dec 17, 2013 at 05:34:44PM +0100, Jakub Hrozek wrote:
On Tue, Dec 17, 2013 at 05:02:07PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 10:09:58PM +0100, Sumit Bose wrote:
On Mon, Dec 16, 2013 at 07:05:16PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 07:03:11PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote:
On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote: > Hi, > > I found this bug when testing the GC patches. Previously, when SSSD was > started, but subdomains list was up-to-date, the ad_ctx was not > initialized for the subdomain. > > I was also thinking whether we should re-initialize the domain-realm > mappings after sssd startup, the way we re-initialize kdcinfo files. I > don't think it's strictly necessary because if someone deletes a file in > /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we > should be at least aware..
Maybe we can do this during the init phase of the responder? We
ah, sorry, I had the typo here ^^^^^^^^ first. I meant to call it during the initialization of the subdomain part of the AD and IPA providers.
bye, Sumit
That makes perfect sense :-)
See the attached patch, I moved the code to a separate function.
Actually, I did one more change in the original patch and also did a similar change in the IPA provider. The result is a second patch. I mostly split the patches into two because the IPA patch is a convenience, so from my point of view it's fine to fix it in master only. The first one is something I'd like in sssd-1-11, too.
ACK to both. I do not see a reason why the IPA patch should not be pushed to 1.11.
bye, Sumit
On Wed, Dec 18, 2013 at 02:58:17PM +0100, Sumit Bose wrote:
On Tue, Dec 17, 2013 at 05:34:44PM +0100, Jakub Hrozek wrote:
On Tue, Dec 17, 2013 at 05:02:07PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 10:09:58PM +0100, Sumit Bose wrote:
On Mon, Dec 16, 2013 at 07:05:16PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 07:03:11PM +0100, Jakub Hrozek wrote:
On Mon, Dec 16, 2013 at 12:04:44PM +0100, Sumit Bose wrote: > On Sat, Dec 14, 2013 at 10:18:32PM +0100, Jakub Hrozek wrote: > > Hi, > > > > I found this bug when testing the GC patches. Previously, when SSSD was > > started, but subdomains list was up-to-date, the ad_ctx was not > > initialized for the subdomain. > > > > I was also thinking whether we should re-initialize the domain-realm > > mappings after sssd startup, the way we re-initialize kdcinfo files. I > > don't think it's strictly necessary because if someone deletes a file in > > /var/lib/sss/pubconf, he should keep the broken pieces, but perhaps we > > should be at least aware.. > > Maybe we can do this during the init phase of the responder? We
ah, sorry, I had the typo here ^^^^^^^^ first. I meant to call it during the initialization of the subdomain part of the AD and IPA providers.
bye, Sumit
That makes perfect sense :-)
See the attached patch, I moved the code to a separate function.
Actually, I did one more change in the original patch and also did a similar change in the IPA provider. The result is a second patch. I mostly split the patches into two because the IPA patch is a convenience, so from my point of view it's fine to fix it in master only. The first one is something I'd like in sssd-1-11, too.
ACK to both. I do not see a reason why the IPA patch should not be pushed to 1.11.
bye, Sumit
Thanks for the review, pushed to master and sssd-1-11
sssd-devel@lists.fedorahosted.org
-
Jakub Hrozek -
Sumit Bose