My Oracle Support Banner

Dm_ifw_sync Startup Fails With Error "Bad Mmap(2) Errno: 12" (Doc ID 1601869.1)

Last updated on FEBRUARY 18, 2019

Applies to:

Oracle Communications Billing and Revenue Management - Version 7.2.0.3.0 to 7.5.0.0.0 [Release 7.2.0 to 7.5.0]
Information in this document applies to any platform.

Purpose

Some concerns about the shared memory errors while starting DM are addressed here.

Questions and Answers

To view full details, sign in with your My Oracle Support account.

Don't have a My Oracle Support account? Click to get started!


In this Document
Purpose
Questions and Answers
 Qn1: On BRM 7.2.0.3.0 version, dm_ifw_sync fails to start with error "bad mmap(2) errno: 12 "
 Qn2: Is the function mmap() using MAP_FIXED ? If so, what is the parameter/value in pin.conf that will be checked as fixed address ? Is the source code first getting the shared memory with shmget() and then mapping the allocated shared memory segment with a file using mmap() ? Because we were confused, as to when the shared memory details were asked, as mmap() will use the process memory right ?
 Qn3: In the output of "ipcs -ma" command, the shared segments have the Creator user id as root against the DM process. But we start the DM_ORACLE or DM_IFW_SYNC processes with pin user. Our understanding is that the shared memory segments should have the Creator user id as pin.
 Qn4: Since all processes are sharing the global shared memory region, is it possible to know the maximum contiguous memory segment possible at any moment. Or do we have any utilities to tackle the fragmentation?
 Qn5: In regards to the performance, our understanding is that bill run will take less time, when the shared memory chunk allocated is huge. Is it not the case ?
 Qn6 : We see a discrepancy in the error printed in dm log file while starting.
 Qn7: We have used 107MB per DM. With 10 DM_ORACLE processes and 1 DM_IFW_SYNC process, the total will come as 1234173952 bytes. But this was not matching with the ipcs -ma output.
 Qn8: We used to have 118M as shared memory for the bill run. Now we are going ahead with 107M for the bill run. But in production, it has to bill approx 30Million subscribers. So is there any chance of water mark error coming up during the billing process ?
 Qn9: If high water mark (HWM) error comes for DM, that is, the heap is used up, and following that if 100% is used up, then will the DM process die? If so what is the impact on the bill run?
 Qn10:  Below is the DM log information for one of the DM process after sending USR1 signal. Does it show that we have enough heap ?
References


My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.