| Title: | DEC TCP/IP Services for OpenVMS |
| Notice: | Note 2-SSB Kits, 3-FT Kits, 4-Patch Info, 7-QAR System |
| Moderator: | ucxaxp.ucx.lkg.dec.com::TIBBERT |
| Created: | Thu Nov 17 1994 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 5568 |
| Total number of notes: | 21492 |
Hi,
A customer of us report intermittent hangs in telnet sessions.
We would like to know if anybody has seen this behavior before or to
receive any suggestion about how to find the origin of the problem.
First, here it's some data about the customer configuration:
UCX> show version
Digital TCP/IP Services for OpenVMS Alpha Version V4.1 - ECO Level 3
on a AlphaServer 8400 Model EV56/440 running OpenVMS V6.2-1H3
UCX> show service telnet/full
Service: TELNET
State: Enabled
Port: 23 Protocol: TCP Address: 0.0.0.0
Inactivity: 1 User_name: not defined Process: not defined
Limit: 1500 Active: 467 Peak: 520
File: not defined
Flags: Listen Rtty
Socket Opts: Keepalive Rcheck Scheck
Receive: 3000 Send: 3000
Log Opts: Actv Dactv Conn Error Logi Logo Mdfy Rjct
File: not defined
Security
Reject msg: not defined
Accept host: 0.0.0.0
Accept netw: 0.0.0.0
UCX> show inter/full
Interface: WE0
IP_Addr: 3.3.0.13 NETWRK: 255.0.0.0 BRDCST: 3.255.255.255
Ethernet_Addr: AA-00-04-00-0D-0C MTU: 1500
Flags: UP BRDCST NOTRL RUN
RECEIVE SEND
Packets 13053387 12303602
IP packets 12421551 11776104
Broadcast IP packets 25929 0
Trailer 1 IP packets 0
Trailer 2 IP packets 0
ARP packets 523562 3352
Broadcast ARP packets 82345 524143
Drops
IP packets 0 0
ARP packets 0 0
Errors
Hardware 0 0
Software 0 0
Restarting attempts 0
Successful restarts 0
Interface: WF0
IP_Addr: 11.0.0.13 NETWRK: 255.0.0.0 BRDCST: 11.255.255.255
Ethernet_Addr: 00-00-F8-45-B2-8B MTU: 4348
Flags: UP BRDCST NOTRL RUN
RECEIVE SEND
Packets 41795 38881
IP packets 40036 37309
Broadcast IP packets 0 0
Trailer 1 IP packets 0
Trailer 2 IP packets 0
ARP packets 1481 87
Broadcast ARP packets 278 1485
Drops
IP packets 0 0
ARP packets 0 0
Errors
Hardware 0 0
Software 0 0
Restarting attempts 0
Successful restarts 0
Interface: LO0
IP_Addr: 127.0.0.1 NETWRK: 255.0.0.0 BRDCST:
Ethernet_Addr: MTU: 65535
Flags: UP LOOP
RECEIVE SEND
Packets 1689 11335
IP packets 1689 11335
Broadcast IP packets 0 0
Trailer 1 IP packets 0
Trailer 2 IP packets 0
ARP packets 0 0
Broadcast ARP packets 0 0
Drops
IP packets 0 0
ARP packets 0 0
Errors
Hardware 0 0
Software 0 0
Restarting attempts 0
Successful restarts 0
UCX> show config inter/full
Interface: WE0
IP_Addr: 3.3.0.13 NETWRK: 255.0.0.0 BRDCST: 3.255.255.255
C_Addr: C_NETWRK: C_BRDCST:
Flags:
Receive buffer: 0
Interface: WF0
IP_Addr: 11.0.0.13 NETWRK: 255.0.0.0 BRDCST: 11.255.255.255
C_Addr: C_NETWRK: C_BRDCST:
Flags:
Receive buffer: 0
Interface: LO0
IP_Addr: 127.0.0.1 NETWRK: 255.0.0.0 BRDCST:
UCX> show comm
Communication Parameters
Local host: prove1 Domain: tm.es
Cluster timer: 5
Maximum Current Peak
Interfaces 20 3 3
Device_sockets 4000 2099 2118
Routes 65535 6 6
Services 1000 494 521
Proxies 20
Type: FDDI Free Maximum Max Bytes Minimum Min Bytes
Large buffers 20 900 4262400 10 47360
Small buffers 150 7500 1920000 50 12800
IRPs 20 2000
Non UCX buffers 10
Remote Terminal
Large buffers: 10
UCBs: 4
Virtual term: disabled
UCX> show config comm
Communication Configuration
Local host: prove1 Domain: tm.es
Cluster timer: 0
Interfaces: 0 Type: Default
Device_sockets: 4000
Routes: 0
Services: 1000
Proxies: 0
Free Maximum Minimum
Large buffers 0 900 0
Small buffers 0 7500 0
IRPs 0 2000
Non UCX buffers 0
Remote Terminal
Large buffers: 0
UCBs: 0
Virtual term: disabled
$ show cpu
PROVE1, a AlphaServer 8400 Model EV56/440
Multiprocessing is ENABLED. Streamlined synchronization image loaded.
Minimum multiprocessing revision levels: CPU = 1
PRIMARY CPU = 00
Active CPUs: 00 01 02 03 04 05 06 07 08 09
Configured CPUs: 00 01 02 03 04 05 06 07 08 09
$
$ show mem
System Memory Resources on 5-MAR-1997 12:55:43.88
Physical Memory Usage (pages): Total Free In Use Modified
Main Memory (6144.00Mb) 786432 304321 417122 64989
Virtual I/O Cache (Kbytes): Total Free In Use
Cache Memory 3200 0 3200
Granularity Hint Regions (pages): Total Free In Use Released
Execlet code region 1024 0 810 214
Execlet data region 256 6 186 64
VMS exec data region 17744 0 17744 0
Resident image code region 1024 213 811 0
Slot Usage (slots): Total Free Resident Swapped
Process Entry Slots 1748 511 1237 0
Balance Set Slots 1746 511 1235 0
Dynamic Memory Usage (bytes): Total Free In Use Largest
Nonpaged Dynamic Memory 119996416 64279040 55717376 63299264
Paged Dynamic Memory 49995776 8311840 41683936 8110192
Paging File Usage (blocks): Free Reservable Total
SYS_PAGE_1:[PAGE]SWAPFILE_1.SYS;1 19968 19968 19968
SYS_PAGE_2:[PAGE]SWAPFILE_2.SYS;1 19968 19968 19968
SYS_PAGE_1:[PAGE]PAGEFILE_1.SYS;1 1806304 1154064 1999872
SYS_PAGE_2:[PAGE]PAGEFILE_2.SYS;1 1799856 1153232 1999872
SYS_PAGE_3:[PAGE]PAGEFILE_3.SYS;1 1809968 1156560 1999872
SYS_PAGE_4:[PAGE]PAGEFILE_4.SYS;1 1819824 1155232 1999872
SYS_PAGE_5:[PAGE]PAGEFILE_5.SYS;1 1805984 1155184 1999872
SYS_PAGE_6:[PAGE]PAGEFILE_6.SYS;1 1815680 1153152 1999872
SYS_PAGE_1:[PAGE]PAGEFILE_7.SYS;1 1821248 1155568 1999872
SYS_PAGE_2:[PAGE]PAGEFILE_8.SYS;1 1838704 1154496 1999872
SYS_PAGE_3:[PAGE]PAGEFILE_9.SYS;1 1829616 1156720 1999872
SYS_PAGE_4:[PAGE]PAGEFILE_10.SYS;1 1826656 1154928 1999872
SYS_PAGE_5:[PAGE]PAGEFILE_11.SYS;1 1840576 1155408 1999872
SYS_PAGE_6:[PAGE]PAGEFILE_12.SYS;1 1838352 1154880 1999872
Of the physical pages in use, 24924 pages are permanently allocated to OpenVMS.
==========End of Config============================
Here it's a more detailed description about the behavior of this
hangs and how we can reproduce it:
1.- Working with the system trough a telnet session, suddenly the
session hang and after some minutes (3 � 4), the session resume its normal
working showing the result of any data o command done. During the hang it's
posible to open a new telnet session and to work normally.
This behaviour has not seen working with decnet sessions ($set host).
We can reproduce the problem at customer site openning a telnet
session a repeating any dcl command (ex:$show sys) a lot of times.
Here it's some date about the test we have done:
1.- With telnet session through the ethernet interface the problem
arise very soon and the hang are frecuents.
2.- With telnet session through the FDDI interface is more difficult
to get the hang but it occurs.
3.- With telnet 0 session at system console the hang occurs too.
4.- With decnet session ($set host) we have been unable to reproduce
the problem.
5.- Working at system console we have been unable to reproduce the
problem.
Any idea?
Ventura Acera
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 5299.1 | Try to check this | VIRGIN::BASSI | Thu Mar 06 1997 03:02 | 98 | |
Hello Ventura,
assuming you are experiencing TELNET-outgoing connect requests problems only
try to check the following:
* Enough local System ressources?
-------------------------------
- NPAGEDYN
- PAGEDYN
- CHANNELCNT
- MAXPROCESSCNT
- VIRTUALPAGECNT
- Did you run "SYS$UPDATE:AUTOGEN GENPARAMS SAVPARAMS" and checked
SYS$SYSTEM:AGEN$PARAMS.REPORT ?
- $ SET LOGIN/INTER=xxx
- Enough Licenses units
* Enough local UCX ressources ?
-----------------------------
- $ UCX SHOW COMM/MEM (check for waits and drops)
- Enough UCX$INET_ACP process quotas (check with SDA; if to few quotas
then edit SYS$MANAGER:UCX$INET_STARTUP.COM and increment related
Process Quotas (This is for incoming only!!!)
* Is local ARP Cache size big enough ?
--------------------------------------
- $ UCX SHOW PROTOCOL ARP/PARAM
* What's about TCP statistics?
------------------------------
- $ UCX SHOW PROTOCOL TCP
* What's about the IP statistics?
-------------------------------
- $ UCX SHOW PROTOCOL IP
* what's about network topology between local UCX-Host and target host(s)?
-------------------------------------------------------------------------
- Bottlenecks for IP ?
- Routing configuration problems ?
- Do (eventual) routers have enough ressources ?
* What's about Data Link traffic?
---------------------------------
- $ MCR NCP SHOW KNOWN LINE COUNTER
- $ MCR NCP SHOW KNOWN CIRCUIt COUNTER
- $ MCR NCL SHOW CSMA-CD STATION <stat-name> ALL COUNTER
- $ MCR NCL SHOW FDDI STATION <stat-name> ALL COUNTER
- Dif you collect some TCP or IP traces during TELNET session
establishement ?
- Did you analyze the network ?
* How does ity look like on the remote TELNET-srv node(s)?
----------------------------------------------------------
- Fast enough to handle many many TELNET
incoming connect request ?
- Still enough system ressources ?
- If not Unix, still enough TELNET ressources ?
UCX V4.1-AXP -ECO4 release notes do not mention anything about your
outgoing TELNET pbs
Since everything above seems to be O.K., try to contact UCX-Engineering.
Hope this will help a bit.
Hasta la vista.
gianbattista
| |||||
| 5299.2 | TELNET Incomming | MDR01::VENTURA | MCS Madrid | Thu Mar 06 1997 10:39 | 148 |
Hello Gianbattista,
Thanks for your quick answer, your tips are very useful, but the
problem we are experiencing are Telnet-incoming sessions.
The problem is with sessions openned in a localized system (the one
of .0 with 10 cpu's and 6GB of main memory) from a lot of different sources
(pc's, unix, vms-ucx).
The network configuration doesn't affect becouse the customer can
reproduce the problem doing a 'telnet 0' session.
Below are some more data about ucx and the ucx$inet_acp proccess, but
I don't see anything wrong.
Regards
Ventura Acera
----------------------------
$ucx show comm/mem
MBUF Summary
Small_static Large_static Small_dynamic Large_dynamic
Total buffers 50 10 3350 30
Free 0 0 53 4
Busy
Data 0 10 0 26
Header 0 0 869 0
Socket 29 0 1210 0
Prot. control 9 0 1218 0
Route 6 0 0 0
Socket name 0 0 0 0
Socket options 0 0 0 0
Fragment reassembly 0 0 0 0
IP address 3 0 0 0
Size of cluster 13056 47616 30288 12112
Free Current Peak Waits Drops
Small Buffers 3347 5721 0 0
Large Buffers 36 168 0 0
IRPs 5 822 1061 0 0
Small clusters Large clusters Non UCX buffers
Free 0 1 0
$ ucx show prot arp /param
ARP
Cache size: 0
Complete timer: 20
Incomplete timer: 3
$
UCX> show protocol tcp
TCP
Connect initiated: 4983 Connect accepted: 24456
Connect established: 29354 Connect closed: 37630
Connect dropped: 10305 Embry connect drop: 1746
Attempt rtt: 20160858 Succeeded rtt: 20120662
XMT Delayed ACKs: 2132695 Connect timeout: 117
ReXMT timeout: 69317 Persist timeout: 587
Keepalive timeout: 459230 Keepalive probes: 390599
Keepalive drops: 165 Total XMT segments: 26697852
XMT segments: 23703583 XMT bytes: 3502703187
XMT packet reXMT: 69710 XMT bytes reXMT: 13192016
XMT ACK only: 2704033 XMT window probes: 495
XMT URG only: 0 XMT wind update pack: 182200
XMT CTRL segments: 37831 Total RCV segments: 25654876
RCV segments: 16642201 RCV bytes: 1869681030
RCV chksum error: 59 RCV bad offset: 0
RCV too short: 0 RCV dup only pack: 547062
RCV dup only bytes: 47381438 RCV part dup pack: 3749
RCV part dup bytes: 38230 RCV bad order pack: 11250
RCV bad order bytes: 19771 RCV pack after wind: 0
RCV bytes after wind: 0 RCV pack after close: 240
RCV window probes: 0 RCV dup ACKs: 386071
RCV ACK for unXMT: 0 RCV ACK segments: 20534734
RCV ACK bytes: 3503248443 RCV wind update pack: 36453
UCX>
UCX> show prot ip
IP
Bad IP Version drops: 0
Unknown IP broadcast address drops: 10098
Total datagrams: 34284076
Datagrams with bad header checksums: 0
Datagrams with size smaller than minimum: 7
Datagrams with data size < data length: 0
Datagrams with header length < data size: 0
Datagrams with data length < header length: 0
Datagrams forwarded: 0
Fragments received: 0
Fragments dropped: 0
Fragments timed out: 0
Datagrams received for unreachable destination: 0
Datagrams forwarded on same net: 0
Fragments dequeued from reassembly queue: 0
Fragments queued in reassembly queue: 0
Datagrams dropped during reassembly: 0
Datagrams successfully reassembled: 0
Hosts IP address access list drops: 0
Networks IP address access list drops: 0
SDA> sh proc
Process index: 001B Name: UCX$INET_ACP Extended PID: 0000081B
-----------------------------------------------------------------
Process status: 00958011 RES,PSWAPM,NOACNT,PHDRES,LOGIN,NODELET
Required capabilities: 0000000C QUORUM,RUN
PCB address 82BA7CC0 JIB address 828E5700
PHD address A7234000 Swapfile disk address 00000000
Master internal PID 0001001B Subprocess count 0
Internal PID 0001001B Creator internal PID 00000000
Extended PID 0000081B Creator extended PID 00000000
State HIB Termination mailbox 0000
Previous CPU Id 00000002 Current CPU Id 00000002
Previous ASNSEQ 000000000006468A Previous ASN 000000000000002A
Current priority 13 # of threads 0000000000000000
Initial process priority 8 Delete pending count 0
Base priority 8 AST's active NONE
UIC [00001,000004] AST's remaining 509
Mutex count 0 Buffered I/O count/limit 32767/32767
Waiting EF cluster 0 Direct I/O count/limit 32767/32767
Abs time of last event 00BA2C98 BUFIO byte count/limit 5395328/5395328
Event flag wait mask A6484AB0 # open files allowed left 99
Process index: 001B Name: UCX$INET_ACP Extended PID: 0000081B
-----------------------------------------------------------------
Swapped copy of LEFC0 00000000 Timer entries allowed left 15
Swapped copy of LEFC1 00000000 Active page table count 0
Global cluster 2 pointer 00000000 Process WS page count 81
Global cluster 3 pointer 00000000 Global WS page count 33
SDA>
$ ! + ( 4 * LRPSIZE) for ARP )
$ !
$ RUN 'UCX$ACP_IMAGE' -
/NOACCOUNTING -
/NOAUTHORIZE -
/NOSWAPPING -
/AST_LIMIT=512 -
/BUFFER_LIMIT=6000000 -
/FILE_LIMIT=10 -
/IO_BUFFERED=32767 -
/IO_DIRECT=32767 -
/QUEUE_LIMIT=16 -
/ENQUEUE_LIMIT=100 -
/PRIORITY=8 -
/PRIVILEGES=CMKRNL -
/PROCESS_NAME='UCX$ACP_NAME' -
/UIC=[1,4]
$ !
$ WRITE SYS$OUTPUT "The Internet driver and ACP were successfully loaded."
| |||||
| 5299.3 | UCXAXP::GEMIGNANI | Fri Mar 07 1997 18:16 | 12 | ||
First of all, can I have that configuration set up right here in my
office? It's got 6GB of memory and enough pagefile space to store the
entire Christmas List of ALL THREE of my children.
On the serious side, there has been some work on quota problems around
RSH. I didn't notice if your UCX was up-to-date or not. Earlier
releases have a bug where the AST quotas run out in INETACP and, as a
result, certain connections no longer get accepted. If you aren't at
the latest ECO level, please get and apply the ECO. If you are, then
please submit an IPMT case with as much information as you can muster
(please add the quota information of the INETACP, possibly output from
SDA).
| |||||