Re: [Freeipa-devel] [PATCH 0038] Reduced time for IO blocking of DS
On 02.06.2016 10:09, thierry bordaz wrote: On 06/02/2016 09:48 AM, Martin Basti wrote: On 31.05.2016 17:10, Stanislav Laznicka wrote: Hello, This is a fix to https://fedorahosted.org/freeipa/ticket/5383. From the comments I am not sure if nsslapd-idletimeout should be reduced as well. If so, could you please propose a value that you find reasonable? Thanks, Standa It looks that only ioblocktimeout is safe to change and we should not change the second attribute, Thierry can you confirm this? Martin^2 The patch looks ok to me. It is safe to tune idletimeout as well but here it does not bring real benefit. The patch leaves idletimeout to the default value that is no close on idle connection. ioblock is set to 10s that is in a good range of values (DS team recommended 10-30s). Note that ioblock is sensitive for slow ldap client. If a client is not able to read fast enough, the server will close the connection. This can happen for example if a client did a large SRCH and is processing slowly the received entries. thierry ACK, works for me Pushed to master: e9f0e9d8fafb35acf0adbd99d40c16edf510ff9f -- Manage your subscription for the Freeipa-devel mailing list: https://www.redhat.com/mailman/listinfo/freeipa-devel Contribute to FreeIPA: http://www.freeipa.org/page/Contribute/Code
Re: [Freeipa-devel] [PATCH 0038] Reduced time for IO blocking of DS
On 06/02/2016 09:48 AM, Martin Basti wrote: On 31.05.2016 17:10, Stanislav Laznicka wrote: Hello, This is a fix to https://fedorahosted.org/freeipa/ticket/5383. From the comments I am not sure if nsslapd-idletimeout should be reduced as well. If so, could you please propose a value that you find reasonable? Thanks, Standa It looks that only ioblocktimeout is safe to change and we should not change the second attribute, Thierry can you confirm this? Martin^2 The patch looks ok to me. It is safe to tune idletimeout as well but here it does not bring real benefit. The patch leaves idletimeout to the default value that is no close on idle connection. ioblock is set to 10s that is in a good range of values (DS team recommended 10-30s). Note that ioblock is sensitive for slow ldap client. If a client is not able to read fast enough, the server will close the connection. This can happen for example if a client did a large SRCH and is processing slowly the received entries. thierry -- Manage your subscription for the Freeipa-devel mailing list: https://www.redhat.com/mailman/listinfo/freeipa-devel Contribute to FreeIPA: http://www.freeipa.org/page/Contribute/Code
Re: [Freeipa-devel] [PATCH 0038] Reduced time for IO blocking of DS
On 31.05.2016 17:10, Stanislav Laznicka wrote: Hello, This is a fix to https://fedorahosted.org/freeipa/ticket/5383. From the comments I am not sure if nsslapd-idletimeout should be reduced as well. If so, could you please propose a value that you find reasonable? Thanks, Standa It looks that only ioblocktimeout is safe to change and we should not change the second attribute, Thierry can you confirm this? Martin^2 -- Manage your subscription for the Freeipa-devel mailing list: https://www.redhat.com/mailman/listinfo/freeipa-devel Contribute to FreeIPA: http://www.freeipa.org/page/Contribute/Code