KNOWN ISSUE: TFTP downloads hang when there is a network interuption

Products

• Deployment Solution

Issue/Introduction

When a client is downloading Linux or WinPE over TFTP and there is a network interuption that would cause a retry to occur, the TFTP times-out.

The MTFTP log shows:

D [11:18:19 04/02] (6060): HandleTFTP: action: next block  IPAddr=192.168.16.20:2073
  (6060)HandleTFTP: P_ACK for blk: 78 want blk: 79
  (6060)SendUCast: blk = 79  IPAddr: 192.168.16.20
  (6060)Cache 0 Data Returned (refread:1)
  (6060)MySendTo completed: len = 772
D [11:18:19 04/02] (6060): HandleTFTP: action: next block  IPAddr=192.168.16.20:2073
  (6060)HandleTFTP: P_ACK for blk: 79 want blk: 80
  (6060)SendUCast: blk = 80  IPAddr: 192.168.16.20
  (6060)Cache 0 Data Returned (refread:1)
  (6060)MySendTo completed: len = 772
D [11:18:19 04/02] (4880): InUseStatus: >>- MTFTP Service  (1238696088)  status=In use  exit=False
D [11:18:20 04/02] (6060): Retry TimeOut: FHd: 1  IPAddr=192.168.16.20
D [11:18:20 04/02] (6060): Enter: RetryTimeout(...)
  (6060)Cache 0 Data Returned (refread:1)
  (6060)MySendTo completed: len = 772
  (6060)PxeMtftp: RetryTimeout()
  (6060)     File Name: d:\Altiris\eXpress\Deployment Server\PXE\images\MenuOption131\X86PC\bootmgr.exe
  (6060)      File Idx: 1
  (6060)        IPAddr: 192.168.16.20
  (6060)          Port: 2073
  (6060)    Last Block: 81
D [11:18:21 04/02] (6060): Retry TimeOut: FHd: 1  IPAddr=192.168.16.20
D [11:18:21 04/02] (6060): Enter: RetryTimeout(...)
  (6060)Cache 0 Data Returned (refread:1)
  (6060)MySendTo completed: len = 772
  (6060)PxeMtftp: RetryTimeout()
  (6060)     File Name: d:\Altiris\eXpress\Deployment Server\PXE\images\MenuOption131\X86PC\bootmgr.exe
  (6060)      File Idx: 1
  (6060)        IPAddr: 192.168.16.20
  (6060)          Port: 2073
  (6060)    Last Block: 81

The last 2 message will repeat until you see

D [11:19:35 04/02] (4880): InUseStatus: >>- MTFTP Service  (1238696088)  status=Not in use  exit=False
I [11:20:33 04/02] (672): Enter: PxeMTFTP::StopService(...)
I [11:20:33 04/02] (592): Enter: PxeMtftp::Terminate(...)

In Wireshark you will see the following:

Note that on block: 555 the server has started using 192.168.1.11 instead of using 192.168.16.11.  The block number will not always be 555, but the behavior of switching IPs is.

Cause

There is a defect in the MTFTP server code that causes all retries to default to the first NIC on the Deployment Server.

Resolution

Development is aware of this issue.  Currently there is no ETA on the fix.

There are two possible workarounds to this issue:

Workaround 1:  Put a PXE server on each VLAN and use PXE forced mode to have those VLANs point to their respective PXE server.  This can even be done by using virtual machines on the Deployment Server.

Workaround 2:  Use a third party TFTP server.  Details on this can be found at https://www-secure.symantec.com/connect/articles/double-your-pxe-boot-speed.

Applies To
DS 6.9 SP1

This only occurs on Deployment Servers with multiple interfaces serving PXE