replset9.js has failed several times recently on the Linux RHEL 32-bit builder:

Build #1636 on June 29

• http://buildlogs.mongodb.org/Linux%20RHEL%2032-bit/builds/1636/test/replica%20sets/replset9.js
• "Sun Jun 30 01:14:52.183 [awaiting replication] timed out after 640000ms ( 321 tries ) at src/mongo/shell/utils.js:492"

Build 1639 on July 2

• http://buildlogs.mongodb.org/Linux%20RHEL%2032-bit/builds/1639/test/replica%20sets/replset9.js
• "Tue Jul 2 01:14:14.091 [awaiting replication] timed out after 640000ms ( 321 tries ) at src/mongo/shell/utils.js:492
  failed to load: /data/buildslaves/Linux_RHEL_32bit/mongo/jstests/replsets/replset9.js"

Build 1642 on July 3

• http://buildlogs.mongodb.org/Linux%20RHEL%2032-bit/builds/1642/test/replica%20sets/replset9.js
• "Wed Jul 3 10:03:12.938 [awaiting replication] timed out after 640000ms ( 321 tries ) at src/mongo/shell/utils.js:492"

Previous failures of this test on Windows and FreeBSD were described in SERVER-9623 and SERVER-7701. In these cases the answer seemed to be "box is slow, so we made the test faster"

I don't see evidence of this occurring on the Linux RHEL 32-bit builder before June 26 (rev bd0713e9a67f)

http://buildbot.mongodb.org/builders/Linux%20RHEL%2032-bit?numbuilds=100

Perhaps we introduced a performance regression that is causing this particular canary to start chirping?

Eric - assigning to you for triage