From: Sage Weil <sage@redhat.com>
Date: Tue, 10 Sep 2019 21:41:18 +0000 (-0500)
Subject: osdc/Objecter: resend OSD tell commands on EAGAIN
X-Git-Tag: v15.1.0~1323^2~24
X-Git-Url: http://git.apps.os.sepia.ceph.com/?a=commitdiff_plain;h=ecfd5ba4cf8dc265620027cd0a8fcd3b2a962d12;p=ceph.git

osdc/Objecter: resend OSD tell commands on EAGAIN

Request map *and* resend.  We don't have map epoch info about when the
reply was sent, and the OSD isn't ordering with respect to epochs anyway.
So, resend now, just in case we already saw a map change, or because we
were suffering from a peering vs command race on the OSD side, and then
also request a new map from the mon, in case we are missing a map update.

Signed-off-by: Sage Weil <sage@redhat.com>
---

diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index b733173fd276c..aa38e1f82cdf0 100644
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -2411,7 +2411,6 @@ void OSD::asok_command(
 	ss << "not primary for pgid " << pgid;
 	// do not reply; they will get newer maps and realize they
 	// need to resend.
-#warning fixme on client side
 	pg->unlock();
 	ret = -EAGAIN;
 	goto out;
diff --git a/src/osdc/Objecter.cc b/src/osdc/Objecter.cc
index a6db93d3dc6e5..6acece251ace5 100644
--- a/src/osdc/Objecter.cc
+++ b/src/osdc/Objecter.cc
@@ -4781,6 +4781,18 @@ void Objecter::handle_command_reply(MCommandReply *m)
     sl.unlock();
     return;
   }
+  if (m->r == -EAGAIN) {
+    ldout(cct,10) << __func__ << " tid " << m->get_tid()
+		  << " got EAGAIN, requesting map and resending" << dendl;
+    // NOTE: This might resend twice... once now, and once again when
+    // we get an updated osdmap and the PG is found to have moved.
+    _maybe_request_map();
+    _send_command(c);
+    m->put();
+    sl.unlock();
+    return;
+  }
+
   if (c->poutbl) {
     c->poutbl->claim(m->get_data());
   }