In the trace record the time value is generated
after acquiring the mutex. The time accuracy is improved
when generated before seizing the mutext.
---
src/base/logtrace.cc| 2 +-
src/base/logtrace_client.cc | 15 ---
src/base/logtrace_client.h | 9 +
3 files
Summary: dtm: improve time accuracy in a trace record [#3144]
Review request for Ticket(s): 3144
Peer Reviewer(s): *** LIST THE TECH REVIEWER(S) / MAINTAINER(S) HERE ***
Pull request to: *** LIST THE PERSON WITH PUSH ACCESS HERE ***
Affected branch(es): develop
Development branch: ticket-3144
Base
Summary: dtm: improve time accuracy in a trace record [#3144]
Review request for Ticket(s): 3144
Peer Reviewer(s): *** LIST THE TECH REVIEWER(S) / MAINTAINER(S) HERE ***
Pull request to: *** LIST THE PERSON WITH PUSH ACCESS HERE ***
Affected branch(es): develop
Development branch: ticket-3144
Base
In the trace record the time value is generated
after acquiring the mutex. The time accuracy is improved
when generated before seizing the mutext.
---
src/base/logtrace.cc| 2 +-
src/base/logtrace_client.cc | 17 ++---
src/base/logtrace_client.h | 9 +
3 files
Hi aThanh,
The CreateLogEntry/... are added recently in LogTraceClient.h/cc, and
the "client" you mean should have been calling the log/trace in
logtrace.h, which are the OpenSAF services and agents. The real client
should include the SAF headers in src/ais/include to use SAF services.
Do
Hello Minh,
1) For LogInternal(..), there is only one version. The new replaced the old.
This is private method, thus it can be safely replaced.
2) For CreateLogEntry(...) which is a public method. For compatibility reasons,
I keep the old method and create the new method. I do not know if
Hi aThanh,
The patch adds a new pair of CreateLogEntry/CreateLogEntryInternal with
one extra parameter. If the old one (within 3 parameters) is not being
used anywhere else, we can delete them.
Thanks
Minh
On 24/1/20 11:34 am, Thanh Nguyen wrote:
In the trace record the time value is
Ack with a comment below:
It is arguable that one second is a big delay interval. How are the other
services handling this?
+ if (error == SA_AIS_ERR_TIMEOUT || error == SA_AIS_ERR_TRY_AGAIN) {
+ redo = true;
+ printf("lockGrantCallbackNoResources received timeout/tryagain\n");
+ sleep(1);