GitLab

Use the correct database entry instead of the frontendDB entry

Performance improvement for indexed attributes removal

Calling from backend.c only works if back-monitor is a static backend

If HAVE_SYS_ERRLIST is false, nothing uses them anyway, so the
declarations serve no purpose. Just breaks the compile if the
library actually defines them (and we are avoiding using them).

Previous commit could cause cmdline cookie to be ignored

If the "" glue entry exists and lacks a contextCSN, must perform
an additional search to be sure the DB is otherwise empty. If so,
skip creating the contextCSN.

Don't resume pending ops unless there are no other threads
waiting to write

Solaris stdio hates NULL pointers

Was overlooked in a0cc1d96

valgrind didn't like accesses to si->si_connaddr
Also fix an array bounds check in ITS#9282 merge_state

Fix a regression in delta-sync, was returning error on old
CSNs instead of ignoring them

Was re-using rid=002, changed to unique rid

Just for convenience

Shows connection address, refresh/persist state, time of last
connect attempt and received data, and last sent and
received cookies per consumer.

Not directly related to syncrepl, but adds a necessary schema arc.
Also add a convenience function for obtaining an entry with known ndn.
Also fix to ignore outbound connections.

Don't log cookiecsn at top, it was already logged on receipt.
Only log the "control csn" and "too old" message once for each sid.

Using cs_pmutex. Reverts the addition of cs_modmutex in ITS#9330,
use cs_pmutex for both delta and plain writes.

Note that plain syncrepl already used cs_pmutex when a cookie CSN
was present in the op. Now it is used for all writes, regardless
of presence of cookie.

The option string, e.g. '<oid>=:dn:' is parsed like a LDIF entry starting from
the '=' and replacing the '=' with a dummy variable 'x'.  In this case, said
string is 'x:dn:', so the resulting effective value is 'dn:'.  This also implies
that base64 values have to be passed in the form '<oid>=::<b64value>'.