When a slave is shut down, it crashes, because we cannot call tdb_down
without tdb_up, and that obviously only happens on a master.
Also, plug a problem inside rtdb_fini (needs a better fix but
should do for now).

Signed-off-by: Pete Zaitcev <zait...@redhat.com>

---
 server/metarep.c |    9 ++++++++-
 server/server.c  |   11 +++--------
 2 files changed, 11 insertions(+), 9 deletions(-)

commit dd6ed2529efa71902f7ac1958f01a2541b4f9133
Author: Pete Zaitcev <zait...@yahoo.com>
Date:   Thu Aug 12 12:27:48 2010 -0600

    Crash on tdb_down if slave.

diff --git a/server/metarep.c b/server/metarep.c
index d3eec49..e13b13f 100644
--- a/server/metarep.c
+++ b/server/metarep.c
@@ -1240,6 +1240,13 @@ int rtdb_restart(struct tablerep *rtdb, bool 
we_are_master)
 void rtdb_fini(struct tablerep *rtdb)
 {
        __rtdb_fini(rtdb);
-       tdb_fini(&rtdb->tdb);
+       /*
+        * This check is ewwww, but unfortunately there's potentially a gap
+        * between DB going master and us bringing up the environment.
+        * If we condition the tdb_fini on DB status, we'll end crashing
+        * if the server terminates during the gap.
+        */
+       if (rtdb->tdb.env)
+               tdb_fini(&rtdb->tdb);
 }
 
diff --git a/server/server.c b/server/server.c
index 8859847..1f8164b 100644
--- a/server/server.c
+++ b/server/server.c
@@ -2326,18 +2326,13 @@ int main (int argc, char *argv[])
        applog(LOG_INFO, "shutting down");
 
        rc = 0;
-
+       if (tabled_srv.state_tdb == ST_TDB_MASTER)
+               tdb_down(&tdbrep.tdb);
        cld_end();
 err_cld_session:
        /* net_close(); */
 err_out_net:
-       if (tabled_srv.state_tdb == ST_TDB_MASTER ||
-           tabled_srv.state_tdb == ST_TDB_SLAVE) {
-               tdb_down(&tdbrep.tdb);
-               rtdb_fini(&tdbrep);
-       } else if (tabled_srv.state_tdb == ST_TDB_OPEN) {
-               rtdb_fini(&tdbrep);
-       }
+       rtdb_fini(&tdbrep);
 err_rtdb:
        event_del(&tabled_srv.pevt);
 err_pevt:
--
To unsubscribe from this list: send the line "unsubscribe hail-devel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to