View Issue Details
ID | Project | Category | View Status | Date Submitted | Last Update |
---|---|---|---|---|---|
0000345 | Other Unix Port | Bug | public | 2009-04-22 12:09 | 2020-08-24 03:13 |
Reporter | rwklein | Assigned To | Steven Levine | ||
Priority | normal | Severity | crash | Reproducibility | random |
Status | closed | Resolution | fixed | ||
Summary | 0000345: STUNNEL 4.26 becomes zombie process | ||||
Description | I have an stunnel session running pretty basic on the mensys mail server with a .conf file and a cerficate. This stunnel runs in server mode. It listens on IP address xx.xx.xx.xx and port x. The decrypted data is send to the localhost and there communigate picks it up. Here in the office runs a local SMTP server, weasel. This one dumps the data to the localhost on the machine running STUNNEL. Thats how the mail gets send to the xxxx SMTP server at IP address xx.xx.xx.xx In some cases the the STUNNEL session on the mailserver turns into a zombie process. The socket on port xx is still open (netstat -l) shows it. But from the other side a telnet to the port no longer shows the welcome text from the mail server. It can also not be killed on the mail server with any process killer. Need to reboot the machine. What happens is netstat -l also shows two figures MAX OVER QUEUED QUEUED The zombie stunnel has overqueded number. Somehow stunnel is drowning in the connections. But its only processing a few email a minute. Steve looked at the code and told me the following: "The stunnel.c at line 264 limits maxclients and maxfd maxfd under libc does not need to be limited." The second bug is when I use Inetmail or weasel to out mail via stunnel. The SMTP demon eats up 100% CPU on the system. Somehow it seems SMTP traffic via stunnel does not work very well. I have no clue why. | ||||
Tags | No tags attached. | ||||
Attached Files | |||||
|
Paul wrote in reply to an email to me: address the line 246 issue This is a legacy from when David from Mensys was maintaining stunnel - perhaps the patch was relevant in the libc05 timeframe when he was looking after the code?? He provided a test version that I have download and will test it. |
|
stunnel-4.34-shl-20110116.zip contains the build that Roderick is currently using. It appears to have cured the SIGSEGVs and has not yet done a silent exit. |
|
The current stunnel build appears to resolve this issue. ftp://ftp.netlabs.org/incoming/stunnel-4.52-shl-20120515.zip |
Date Modified | Username | Field | Change |
---|---|---|---|
2009-04-22 12:09 | rwklein | New Issue | |
2009-04-22 12:19 | rwklein | Note Added: 0001291 | |
2011-02-20 18:20 | Steven Levine | File Added: stunnel-4.34-shl-20110116.zip | |
2011-02-20 18:22 | Steven Levine | Note Added: 0001839 | |
2012-05-25 00:26 | Steven Levine | Note Added: 0002279 | |
2012-05-25 00:26 | Steven Levine | Status | new => resolved |
2012-05-25 00:26 | Steven Levine | Resolution | open => fixed |
2012-05-25 00:26 | Steven Levine | Assigned To | => Steven Levine |
2020-08-24 03:13 | psmedley | Status | resolved => closed |