While applying the October 2026 IDM Stack Patch Bundle (14.1.2.1.261008) to an Oracle Internet Directory (OID) 14c home with SPBAT, the Downtime phase failed with OPatch error code 137. This post explains what that code means, what to check, and the fix that worked for me. The full OID patching walkthrough is in a separate post: [link to your OID walkthrough post].
The symptom
The Downtime phase stopped with:
spbat downtime phase has FAILED.
Patching component oracle.rda, 24.3.24.7.16...
Applying interim patch '39951765' to OH '<ORACLE_HOME>'
<ORACLE_HOME>/OPatch/opatch: line 1365: <pid> Killed $JAVA $otherOpt $JAVA_VM_OPTION ... oracle/opatch/OPatch $args $invptr
OPatch failed with error code 137
Figure 1. The Downtime phase fails: OPatch is killed with error code 137 while applying interim patch 39951765. The pre-checks above it had already passed.
What error code 137 means
Exit code 137 is 128 + 9, which means the process was killed with SIGKILL. That is different from a patch failing on its own terms. The pre-checks had already ended with "All checks passed", so this was not a conflict or a prerequisite problem. Something outside OPatch ended the process, and the word "Killed" in the output is the shell reporting it.
What to check before you rerun
Check 1: Was it the out-of-memory killer?
This is the most common reason for a SIGKILL on a Java process. Look in the kernel log:
dmesg -T | grep -i -E "out of memory|killed process"
grep -i -E "out of memory|killed process" /var/log/messages
Check 2: How much memory and swap are free?
free -m
Check 3: What else is running on the host?
Other Java processes, databases and middleware components all compete for the same memory.
ps aux --sort=-%mem | head
Check 4: Is there a memory limit?
A container or a cgroup limit can kill a process even when the host has free memory.
cat /sys/fs/cgroup/memory.max
ulimit -a
Check 5: How far did the phase get?
Review the SPBAT log and the OPatch log under $ORACLE_HOME/cfgtoollogs/opatch to see which patches were applied before the process was killed.
The fix that worked for me
By default, a JVM chooses its maximum heap size based on the memory of the host. I capped the heap for the OPatch run, then reran the Downtime phase from the SPBAT directory:
export _JAVA_OPTIONS="-Xms512m -Xmx2048m"
echo $ORACLE_HOME
./spbat.sh -type OID -phase Downtime -mw_home $ORACLE_HOME -spb_download_dir /u02/stage/CPUOCTPatch/IDM_SPB_14.1.2.1.261008
In my case, limiting the heap let the rerun complete successfully. _JAVA_OPTIONS applies to every JVM started from that shell, so it limits how much memory OPatch's JVM can ask for. Treat it as a workaround, not a root-cause fix. If the host is regularly short of memory, free up memory by stopping unrelated processes, or add memory, before the next patching window.
Figure 2. The rerun of the Downtime phase after setting _JAVA_OPTIONS completes successfully.
What happened to the patches already applied
The first attempt had already applied some patches before it was killed. The inventory shows patch 39646824 once, with a timestamp from the first attempt, and the remaining patches with later timestamps from the rerun. So the rerun carried on with the patches that were still missing. Before you rerun, check the SPBAT log to see how far the phase got.
$ORACLE_HOME/OPatch/opatch lsinventory | grep -E "Patch|Unique Patch ID"
Figure 3. The inventory after the rerun. The first line, "Picked up _JAVA_OPTIONS", shows the setting was still active in the shell.
Clean up
When patching is finished, unset the variable so it doesn't affect other Java commands in that shell:
unset _JAVA_OPTIONS
What I learned
- Error 137 means killed, not conflicted. Look at the operating system's memory before you look at the patch.
- Check the kernel log first. It tells you quickly whether the out-of-memory killer was involved.
- Capping the JVM heap is a quick workaround when memory is tight, but it doesn't remove the underlying shortage.
- A rerun can resume. The inventory timestamps show what was already applied.
- Unset temporary variables afterwards. They affect every Java command in the session.
Have you seen OPatch exit with 137 on your systems? I'd like to hear what the cause turned out to be.
References
- Patch 40131416, IDM Stack Patch Bundle 14.1.2.1.261008
Kumaran Murugan is an Oracle Apps DBA and OCI Cloud Administrator with 5+ years of experience in Oracle E-Business Suite, database migrations, and OCI infrastructure. Connect on LinkedIn.



Top comments (0)