[Gluster-devel] sles15sp3 gnfs errors about rpc.statd not existing (path change for statd)

[Erik Jacobson]

Dear team -

In our version of glusterfs 9.3, which is the latest we can use due to
internal testing needs, we have made a patch because SLES15 has moved
rpc.statd with no link to the standard location. While we are not using
locking, it was producing error messages. I know this isn't the right
fix for submission but illustrates what we did in a sles15-only build of
glusterfs 9.3. I raced through the release notes and I don't see this
is resolved but I didn't look through git.


diff -Narup glusterfs-9.3.sgi-orig/xlators/nfs/server/src/nlm4.h glusterfs-9.3.sgi/xlators/nfs/server/src/nlm4.h
--- glusterfs-9.3.sgi-orig/xlators/nfs/server/src/nlm4.h	2021-06-29 00:27:44.662408609 -0500
+++ glusterfs-9.3.sgi/xlators/nfs/server/src/nlm4.h	2022-02-12 10:30:55.934953279 -0600
@@ -69,6 +69,10 @@
 #define GF_SM_NOTIFY_PIDFILE "/var/run/sm-notify.pid"
 #endif

+/* sles15 has statd only in /usr/sbin */
+#define GF_RPC_STATD_PROG "/usr/sbin/rpc.statd"
+
+
 extern rpcsvc_program_t *
 nlm4svc_init(xlator_t *nfsx);



PS:
I understand you may not fix this since we are supposed to use Ganesha.
We actually made a lot of progress with Ganesha since I last wrote
about it. Our nfs root node boot problems are better now (the last problem
distro was sles15sp3 but sles15sp4 beta changed something and we don't
hang at boot starting nscd any more!). The reason we haven't switched
yet is a) we still have a supported distro with boot hangs and b)because it's
slower for our workload but we're miles closer than a few months back!! When
we get a chance to breath, we will ask some questions in the Ganesha forum
about performance comparisons gluster NFS to Ganesha with the pynamic library
load simulation (uses mpi) on NFS clients. We have something a lot
closer to a repeatable test case than we had months back. No more daemon
crashes either. Thank you.