Critical semphore errors on a DS16B switch ...

Question

Here is a fun one. Maybe someone can help with it. On a DS16B switch, when the RAPID service is disabled, it will burp out SEMA-GIVE errors and SEMA-TAKE errors immediately when connecting to a switch via telnet. They appear before the login prompt and do not reappear any time else. The switch operates fine and doesn't seem to mind. All of the system services are turned off (rstatd,rusersd,thad, SNMP, Http etc.). Re-enabling RAPID gets rid of the errors. So it looks like the lack of this service is causing the msgs. I tried various levels of firmware and that doesn't help. (It's at current rev now.) Turrning off/on the other services doesn't help either. The other brocade switch models do not have this issue only this hardware model. Any insights would be appreciated. Thanks. Here is a sample ... 0x102bd960 (tShell): Mar 22 08:31:58 CRITICAL SEMA-SEMTAKE, 1, semaTake, sema = 0, errno = 0x3d0001 Traceback: _semaTake+0x344 (0x104f89b4) _ranRscnDeregister+0x1d8 (0x10769ff8) _event_sendSec+0x74 (0x1076a214) _shellTask+0x208 (0x10417188) _vxTaskEntry+0x10 (0x1096ec54) 0x102bd960 (tShell): Fabric OS (tm) Release v3.2.1b login: Mar 22 08:31:58 CRITICAL SEMA-SEMGIVE, 1, semaGive, sema = 0, errno = 0x3d0001 Traceback: _semaGive+0x38 (0x104f8648) _event_sendSec+0x74 (0x1076a214) _shellTask+0x208 (0x10417188) _vxTaskEntry+0x10 (0x1096ec54)

JasonBailey · Accepted Answer

I checked our knowledgbase and no hit.If you raise a case we can escalate it to Brocade.3.x switches run the vxworks platform, 4/5.x run on linux, so that may be why you only see this on 3.x.

sysmgr1 · Answer

I thought this one would be a challenge...

sysmgr1 · Answer

I tried several variations of turning off and on the different system services. None of which solved the problem. It only seems to happen on these 16 port 2G switches. I posted this question on the brocade site. (which is totally useless) Nobody there replied with a fix, only the suggestion to upgrade firmware, which is current and still doesn't work.

sysmgr1 · Answer

This doesn't happen on the 1GB DS16 silkworms either. They are running on the vxworks platform too.

sysmgr1 · Answer

I checked the Knowledgebase (and the rest of the web.) and haven't seen any useful info on the subject. However, the error msgs documentation describe this error as a hardware failure and recommend the switch be replaced. Since this can be easily enabled and disabled by switching the RAPID service on and off, it looks like a software issue.

Opening a case on this isn't worth the enormous hassle it would create. If you want to reproduce this in a lab, setup one of these switches with the latest firmware and then turn off the RAPID service. Logout and then you'll see the messages when you try to log back in (via telnet).

sysmgr1 · Answer

Although it complains and sends out error messages, it appears to not have a negative impact on the switch.

Connectrix

Was this post helpful?