Quanah Gibson-Mount wrote:
--On Friday, October 2, 2020 2:51 PM +0300 openldap-technical(a)kolttonen.fi wrote:
> On Thu, 1 Oct 2020, Quanah Gibson-Mount wrote:
>> Optionally, cd tests && make its to run through the regression suite.
> Second run of "make its" on RHEL8. It seems to me that bdb succeeded this
> time, but now it is hdb that failed its9338:
Ok, thanks. I've run the test through a loop of 100 iterations for both bdb and hdb
on Debian 10 with no issue. I'll test on CentOS8 next. I'm thinking it
may be a timing issue.
Note that this test depends on a socket buffer filling up and causing a slapd
thread to go into a write wait, to even function. It may simply be that the
default kernel socket buffer size on RHEL8 is different and so the test condition
-- Howard Chu
CTO, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/