From: SourceForge.net <no...@so...> - 2003-03-24 09:29:25
|
Bugs item #699319, was opened at 2003-03-07 12:16 You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=112694&aid=699319&group_id=12694 Category: agent Group: None Status: Open Resolution: None Priority: 5 Submitted By: Christian Kleinewaechter (kleinew) Assigned to: Stefan Radman (sradman) Summary: interfaces broken on Solaris with ce-Card and dot1q-Trunking Initial Comment: I have two machines running net-snmp 5.0.7 on Solaris 8 with ce-Interface-Cards and dot1q-Trunking enabled. An snmpwalk on ifTable gives no result while walking interfaces gives only IF-MIB::ifNumber.0: $ snmpwalk mymachine -c mycommunity interfaces IF-MIB::ifNumber.0 = INTEGER: 36 $ snmpwalk mymachine -c mycommunity ifTable $ The symptom is similar to bug 499150. The reason there was that oerror was missing in the ge driver, and when I gather kstat statistics I see instances for the vlan subinterfaces with a fewer entries, while ordinary virtual interfaces have no kstat-entries. $ kstat -c net -m ce module: ce instance: 0 name: ce0 class: net alignment_err 0 brdcstrcv 4895089 brdcstxmt 11251 cap_1000fdx 1 cap_1000hdx 1 cap_100T4 0 cap_100fdx 0 cap_100hdx 0 cap_10fdx 0 cap_10hdx 0 cap_asmpause 0 cap_autoneg 1 cap_pause 0 code_violations 0 collisions 0 crc_err 0 crtime 133.5559658 excessive_collisions 0 first_collision 0 ierrors 0 ifspeed 1000000000 ipackets 651169016 ipackets64 4946136312 ipackets_cpu00 1204252641 late_collisions 0 lb_mode 0 length_err 0 link_T4 0 link_asmpause 0 link_duplex 2 link_pause 0 link_speed 1000 link_up 1 lp_cap_1000fdx 1 lp_cap_1000hdx 0 lp_cap_100T4 0 lp_cap_100fdx 0 lp_cap_100hdx 0 lp_cap_10fdx 0 lp_cap_10hdx 0 lp_cap_asmpause 1 lp_cap_autoneg 1 lp_cap_pause 0 multircv 190 multixmt 0 norcvbuf 0 noxmtbuf 0 obytes 424397976 obytes64 180813024408 oerrors 0 opackets 2312071534 opackets64 2312071534 pci_bad_ack_err 0 pci_drto_err 0 pci_err 0 pci_parity_err 0 pci_rma_err 0 pci_rta_err 0 peak_attempts 0 qos_mode 1 rbytes 893060117 rbytes64 6524948382741 rev_id 17 rx_allocb_fail 0 rx_bad_descs 0 rx_hdr_drops 0 rx_hdr_pkts 28298017 rx_inits 0 rx_len_mm 0 rx_msgdup_fail 0 rx_mtu_drops 0 rx_mtu_pkts 622870999 rx_new_hdr_pgs 884313 rx_new_mtu_pgs 1229459573 rx_new_nxt_pgs 0 rx_new_pages 1230343886 rx_no_buf 9 rx_no_comp_wb 99 rx_nocanput 2071 rx_nxt_drops 0 rx_ov_flow 17 rx_pkts_dropped 0 rx_rel_flow 0 rx_split_pkts 0 snaptime 1799814.8144166 tx_allocb_fail 0 tx_dma_bind_fail 0 tx_inits 0 tx_max_pend 35 tx_msgdup_fail 0 tx_no_desc 0 tx_nocanput 0 tx_queue0 2310147761 tx_queue1 0 tx_queue2 0 tx_queue3 15423 tx_starts 2312071585 tx_uflo 0 xcvr_addr 0 xcvr_id 0 xcvr_inits 1 xcvr_inuse 3 --snip-- module: ce instance: 3000 name: ce3000 class: net brdcstrcv 0 brdcstxmt 6 crtime 135.3945238 ipackets 0 ipackets64 0 maxvids 1024 multircv 0 multixmt 0 nocanput 0 norcvbuf 0 noxmtbuf 0 obytes 276 obytes64 276 opackets 6 opackets64 6 rbytes 0 rbytes64 0 snaptime 1799814.8429867 vid_allocated 34 ---------------------------------------------------------------------- >Comment By: Christian Kleinewaechter (kleinew) Date: 2003-03-24 10:42 Message: Logged In: YES user_id=468236 The ifNumber is correct on both machines. On the first machine which has 1 LO, 1 ERI and 34 CE VLANs, I get ifnumber=36. On the second machine (which was reconfigured meanwhile) I now get ifnumber=6 and have 1 LO, 4 ERIs (1 phys+3 virtual) and 1 CE VLAN. ---------------------------------------------------------------------- Comment By: Stefan Radman (sradman) Date: 2003-03-22 02:48 Message: Logged In: YES user_id=80318 is the ifNumber correct? ---------------------------------------------------------------------- Comment By: Stefan Radman (sradman) Date: 2003-03-20 21:20 Message: Logged In: YES user_id=80318 can you tell me if you get further than ifNumber with the attached patch applied? Stefan ---------------------------------------------------------------------- Comment By: Stefan Radman (sradman) Date: 2003-03-12 17:05 Message: Logged In: YES user_id=80318 Hmmm. It does not explain anything but it's still strange. kstat delivers "ifspeed" as uint64. Not sure if the code can deal with that correctly ... IF-MIB::ifSpeed is declared as Gauge32 --------------------------------------------- $ cat snmpd.log.filtered | grep variable: | uniq | sort -u | grep 'type 4' kernel_sunos5: variable: "ifspeed" (type 4) kernel_sunos5: variable: "ipackets64" (type 4) kernel_sunos5: variable: "obytes64" (type 4) kernel_sunos5: variable: "opackets64" (type 4) kernel_sunos5: variable: "rbytes64" (type 4) ---------- and the ce26000 VLAN interface even delivers the opackets as signed int32 - some might be coming back in again ;-) ---------- kernel_sunos5: module: ce instance: 26000 name: ce26000 class: net type: 1 flags: 0 kernel_sunos5: variable: "ipackets" (type 2) kernel_sunos5: variable: "opackets" (type 1) ---------- tested the same thing on Solaris 7, net-snmp 5.0.7 and an hme interface. ---------- $ snmpd -Lf -CDkernel_sunos5 -c /var/tmp/snmpd.conf 2>&1 | grep ifspeed | head -1 kernel_sunos5: variable: "ifspeed" (type 2) ---------- and here I get back an uint32 (the 64bit counters dont exist) I just dont understand why you dont get ANYTHING except ifNumber. At least I'm able to reproduce the 64bit "ifspeed" on Solaris 8 and 9 with an hme interface and that needs to be fixed first. thanks for the logs. I'll delete the large one. ---------------------------------------------------------------------- Comment By: Christian Kleinewaechter (kleinew) Date: 2003-03-12 14:05 Message: Logged In: YES user_id=468236 Here is kstat.ce ---------------------------------------------------------------------- Comment By: Christian Kleinewaechter (kleinew) Date: 2003-03-12 14:03 Message: Logged In: YES user_id=468236 The file snmpd.log.filtered.gz contains the debugging output without the modules unix, vmem and ssd (the file was to large for upload otherwise). If the complete file is needed I can put it somewhere else for download. kstat.ce the comple kstat. ---------------------------------------------------------------------- Comment By: Stefan Radman (sradman) Date: 2003-03-12 03:55 Message: Logged In: YES user_id=80318 can you run the agent with -Dkernel_sunos5, repeat the snmpwalk over interfaces and attach both output files to this bug? a complete "kstat -c net -m ce" would also be helpful. thanks ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=112694&aid=699319&group_id=12694 |