php.net |  support |  documentation |  report a bug |  advanced search |  search howto |  statistics |  random bug |  login

go to bug id or search bugs for

Bug #57693 Unexpected Memcache::get(): marked server ... as failed
Submitted: 2007-06-06 09:16 UTC Modified: 2007-11-01 09:05 UTC
From: php at koterov dot ru Assigned:
Status: Closed Package: memcache (PECL)
PHP Version: 5.2.1 OS: Unix
Private report: No CVE-ID: None
View Developer Edit
 [2007-06-06 09:16 UTC] php at koterov dot ru 
Description:
------------
Two related descriptions.

1. On heavy loaded server (20-40 requests/s) I periodically (once a second) get an error message from memcached extension:

# tail -f php-error_log |grep failed
PHP Notice:  Memcache::get(): marked server '127.0.0.1:11211' as failed
...

But the server is up, the maximum number of connections (-m 1024) is not reached. And this error does not depend on how long the memcached server is running: a couple of minutes after memcached restart I can watch the same error, the statistics is:

stats
STAT pid 19854
STAT uptime 337
STAT time 1181115051
STAT version 1.2.2
STAT pointer_size 64
STAT rusage_user 1.727737
STAT rusage_system 6.867955
STAT curr_items 63413
STAT total_items 64222
STAT bytes 114838739
STAT curr_connections 16
STAT total_connections 3888
STAT connection_structures 27
STAT cmd_get 259618
STAT cmd_set 64563
STAT get_hits 196161
STAT get_misses 63457
STAT evictions 0
STAT bytes_read 148496793
STAT bytes_written 394115419
STAT limit_maxbytes 1073741824
STAT threads 1

I tried to disable setCompressThreshold, increase memcache.chunk_size,
disable persistent connections, enlarge timeout parameters, but an error
still happens. Furthermore, this error happens on Windows more
frequently, even without heavy load, but I cannot reproduce.

I have the only memcached server added:

$this->connection = new Memcache(); 
$this->connection->addServer(
    "localhost", 
    11211,
    true,
    1,
    10, 
    2
);

2. Seems in some cases I could reproduce this bug: if the key length is wuite large (in my situation - 361 bytes), "marked server '127.0.0.1:11211' as failed" is always reproducible. And this message is generated not immediately, but after the timeout (10 seconds in the example above) - very strange behaviour.

Patches

Pull Requests

History

AllCommentsChangesGit/SVN commitsRelated reports
 [2007-07-10 09:08 UTC] ivyrake at gmail dot com 
I had encountered the same problem,it seem to there have a problem in function php_mmc_store in memcache.c ,it isn't chceck the paramater "key".
if an empty key had been transferred,the function will transfer a wrong command to memcached server.when the error back it call mmc_server_failure(), that change a active server's status to MMC_STATUS_FAILED.

so do a examination in php_mmc_store will resolve that problem.
 [2007-11-01 09:05 UTC] mikael at synd dot info 
This bug has been fixed in CVS.

In case this was a documentation problem, the fix will show up at the
end of next Sunday (CET) on pecl.php.net.

In case this was a pecl.php.net website problem, the change will show
up on the website in short time.
 
Thank you for the report, and for helping us make PECL better.

Fixed problem with some methods modifying keys in-place 
Fixed problem with some methods accepting empty keys, which resulted in server error because an invalid command was sent

Hopefully this will take care of the random problems, you might see some "Keys may not be empty" warnings instead. The client will now return false if a key is empty/invalid.
 
PHP Copyright © 2001-2024 The PHP Group
All rights reserved. 		Last updated: Fri Dec 27 02:01:29 2024 UTC