[SUSE] Crash - Splunk SH is not able to perform a search against indexers.

[sylim_splunk]

We are testing our upgrade from Splunk 7.3.1 to 8.1.2. Once we upgrade the SH regardless of the indexers' version 7.3.1 or 8.1.2, we are unable to run any searches.

Every time we attempt a search, Splunk crashes 100% - crash logs looks like below;

- Resource usage is low as in CPU IDLE 99% memory usages is less than 10%. 

- systemd and bootstart enabled

[build 545206cc9f70] 2021-02-22 17:44:41
Received fatal signal 6 (Aborted).
Cause:
Signal sent by PID 15454 running under UID 1863478.
Crashing thread: RemoteTimelineReadThread
Registers:
RIP: [0x00007F8BE1E67F77] gsignal + 55 (libc.so.6 + 0x34F77)
RDI: [0x0000000000003C5E]

.. Backtrace (PIC build):
[0x00007F8BE1E67F77] gsignal + 55 (libc.so.6 + 0x34F77)
[0x00007F8BE1E6934A] abort + 314 (libc.so.6 + 0x3634A)
[0x000056191C24136E] ? (splunkd + 0x128F36E)
[0x000056191E355416] _ZN10__cxxabiv111__terminateEPFvvE + 6 (splunkd + 0x33A3416)
[0x000056191E355461] ? (splunkd + 0x33A3461)
[0x000056191E355594] ? (splunkd + 0x33A3594)
[0x000056191C2632C6] ? (splunkd + 0x12B12C6)
[0x000056191C263724] ? (splunkd + 0x12B1724)
[0x000056191D28257B] _ZN6ThreadC1EPKcz + 443 (splunkd + 0x22D057B)
[0x000056191DB44181] _ZN39EventDownloadInitializeCollectionThreadC1EP20EventDownloadManager + 33 (splunkd + 0x2B92181)
[0x000056191DB490A8] _ZN20EventDownloadManagerC2ERK3Str + 600 (splunkd + 0x2B970A8)
[0x000056191D7DF4BF] _ZN9TimelinerC2ERK3Str + 319 (splunkd + 0x282D4BF)
[0x000056191C523782] _ZN24RemoteTimelineReadThread4mainEv + 114 (splunkd + 0x1571782)
[0x000056191D281F87] _ZN6Thread8callMainEPv + 135 (splunkd + 0x22CFF87)
[0x00007F8BE21E1724] ? (libpthread.so.0 + 0x8724)
[0x00007F8BE1F1FEED] clone + 109 (libc.so.6 + 0xECEED)
Linux / ito044165 / 4.4.180-94.130-default / #1 SMP Fri Sep 4 09:11:24 UTC 2020 (e5e5142) / x86_64
C++ exception:
exception_addr=0x7f8be158bfc0
typeinfo=0x56191f0bf6e8, name=15ThreadException

what(): RemoteTimelineReadThread: about to throw a ThreadException: pthread_create: Resource temporarily unavailable; 20 threads active

[sylim_splunk · Accepted solution]

The crash happened 100% for any searches when there's no job running or scheduled at all - pthread_create: Resource temporarily unavailable was not triggered by the memory limit.

In search.log, it also has a message as below;

"02-20-2021 17:44:41.364 WARN Thread - RemoteTimelineReadThread: about to throw a ThreadException: pthread_create: Resource temporarily unavailable; 20 threads active"

It was caused by hitting DefaultTasksMax and fixed by adding TasksMax=infinity for the service. Further details found in the link, https://www.suse.com/support/kb/doc/?id=000015901  

[richgalloway]

Does this happen for one particular search or any search?

Check splunkd.log and /var/log/messages for additional information.  Open a support ticket.