*Synopsis*: ksh93 tail -f, with unconditional .25s sleep and line parsing,
about 37x slower than cat
CR 6920072 changed on Mar 30 2010 by <User 1-5Q-1267>
=== Field ============ === New Value ============= === Old Value =============
See Also 6939349
====================== =========================== ===========================
*Change Request ID*: 6920072
*Synopsis*: ksh93 tail -f, with unconditional .25s sleep and line parsing,
about 37x slower than cat
Product: solaris
Category: shell
Subcategory: korn93
Type: Defect
Subtype: Performance
Status: 7-Fix in Progress
Substatus:
Priority: 3-Medium
Introduced In Release: solaris_nevada
Introduced In Build: snv_128
Responsible Engineer: <User 1-5Q-9201>
Keywords: oss-request, oss-sponsor
=== *Description* ============================================================
Running tail -f on a nightly.log during a build on an Opensolaris snv_126
machine,
I noticed that the tail was running long after the build had completed. The
first
time I killed it; the second time I decided to let it run. *26 minutes* after
the
build completed, it finally completed (the entire build had only taken 28
minutes).
Now, this was across an ssh connection from EGO02 to MPK17, but I took the same
log file and did several tests:
tail -f: 26:XX
nfs cp: 00:29
scp: 00:29
scp -C: 00:36
cat: 00:42
That's quite a difference.
Trussing the command, I see that it's ksh93, and that ksh is apparently reading
line-chunked things of approximately 8K, and writing line-chunked blocks to the
output, and unconditionally delaying .25s between each stat/read/write call.
That latter, in particular, seems to be the biggest problem, and seems
ill-advised;
if there's to be a delay, it seems like "stat, and if no change sleep and then
stat again" would be a lot more efficient.
Also, for -f, I don't see why there's any point in doing the line-oriented
calls;
original tail didn't, and I can't imagine it's a win.
*** (#1 of 1): 2010-01-26 04:51:29 GMT+00:00 <User 1-5Q-12546>
=== *Public Comments* ========================================================
=== *Workaround* =============================================================
=== *Additional Details* =====================================================
Targeted Release: solaris_nevada
Commit To Fix In Build: snv_137
Fixed In Build:
Integrated In Build:
Verified In Build:
See Also: 6790507, 6939349
Duplicate of:
Hooks:
Hook1:
Hook2:
Hook3:
Hook4:
Hook5: <email address omitted>
Hook6: <email address omitted>
Program Management:
Root Cause: Another Bug Fix
Fix Affects Documentation: No
Fix Affects Localization: No
=== *History* ================================================================
Date Submitted: 2010-01-26 04:51:29 GMT+00:00
Submitted By: <User 1-5Q-12546>
Status Changed Date Updated Updated By
3-Accepted 2010-03-29 02:46:47 GMT+00:00 <User 1-5Q-9201>
7-Fix in Progress 2010-03-29 03:19:23 GMT+00:00 <User 1-5Q-9201>
=== *Service Request* ========================================================
Impact: Significant
Functionality: Secondary
Severity: 3
Product Name: solaris
Product Release: solaris_nevada
Product Build:
Operating System: snv_126
Hardware: generic
Submitted Date: 2010-01-26 04:51:30 GMT+00:00
=== *Multiple Release (MR) Cluster* - 0 ======================================
