Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

Navigation  Back up to About Overview 
[+] Expand All
[-] Collapse All

No index entries found.

Known Issues

This section lists the known issues in hardware and software in Junos OS Release 15.1F5 for the MX Series and T Series.

For the most complete and latest information about known Junos OS defects, use the Juniper Networks online Junos Problem Report Search application.

General Routing

  • This is a day one issue, kernel and Packet Forwarding Engine are both involved, and the timing factor is playing a role here. This is a single-hop LSP with FRR scenario; the affected destinations are pointing to the unilist next-hop. When the active egress interface flaps very fast, especially if the interface comes back up before kernel gets a chance to delete all the unilist next-hops, those unilist next-hops that have not been deleted yet will be re-used. As a result, the corresponding destinations are pointing to discard next-hop(s) in the Packet Forwarding Engine Jtree. Those unilist next-hops that have been already deleted are safe and get updated accordingly. PR929986
  • Some intermediate hostnames that are displayed incorrectly for the MTR (My traceroute) utility with NAPT44. Traceroute utility works fine. PR1008088
  • On MX Series routers with MPC3E, MPC4E, MPC5E, MPC6E, Junos OS does not support short (sub-second) interface hold-time down configuration. So, a hidden configuartion statement is introduced to ignore DFE tuning state during hold-down timer period. This configuration statement allows sub-second hold-down timer on MPC3E,MPC4E,MPC5E,MPC6E. set interfaces <intf name> hold-time up <U ms> down <D ms> alternative The configuration statement does not work/support 'MPC5E 3D Q 2CGE+4XGE' and 'MIC6 2X100GE CFP2 OTN', and we recommend configuring hold-time down to be more than 3 seconds for these two cards. PR1012365
  • CoS scheduler names cannot be added or changed via service COA's. The schedulers can be added at subscriber login using client dynamic profiles. PR1015616
  • When configuring router in RR mode (cluster-id or option B MP-eBGP peering), the advertise-external feature will not be applicable in local VRFs due to a different route selection/advertisement process (main bgp.l3vpn.0 vs VRF.inet.0). PR1023693
  • The static/static access routes pointing to an unnumbered interface are getting added in the routing table even if the interface is down. In this case, if graceful Routing Engine switchover (GRES) is disabled, this type of routes will never be added in the routing table after Routing Engine switchover. PR1064331
  • SFB2 offline/online with 20 line cards takes 9 minutes 52 seconds whereas for SFB it takes 42 seconds. PR1097338
  • If NSR (nonstop routing) is enabled and a TCP session is terminated while there is still data in the socket pending transmission, the MBUF (kernel memory buffer) used to store this data might not get deallocated properly. In order to hit this issue the TCP session must use NSR active socket replication. If the system runs low on MBUF memory, the kernel will automatically throttle down memory allocation on low priority applications and ultimately, if there is no MBUF left, the system could become unresponsive due to its inability to serve I/O requests. PR1098001
  • On T Series multichassis platforms, when offline and then online the LCC from SCC (e.g. executing the CLI command "set chassis lcc 0 offline" command, and then executing "delete set chassis lcc 0 offline") in quick successions (that is, within the timeout setting for peer to reconnect, 60 seconds, which is not configurable), kernel replication error "ENOENT" may occur, which can cause ksyncd to crash and in thus trigger a live vmcore. Additionally, this is a timing issue and LCC offline followed by online within 60 seconds is the only known trigger so far. As a workaround, on the safer side, it is recommended to online the LCC after 120 seconds. PR1108048
  • On dual Routing Engine MX Series platforms, the "xe" interfaces of any of the line cards the following may flap during in-service software upgrade (ISSU) due to missing support. The flapping may not happen every time and the probability of occurrence would increase if more number of SFP+ (e.g. SFP+-10G-SR) are connected on the FPC. The affected line cards are, * MIC3-3D-10XGE-SFPP * MPC4E-3D-32XGE-SFPP, MPC4E-3D-2CGE-8XGE * MPC5E-40G10G, MPC5EQ-40G10G * MX2K-MIC6-24XE, MX2K-MIC6-24XE-OTN. PR1118379
  • On MS-MPC equipped MX Series platforms, during the "three-way handshake" process, when receiving ACKs (e.g., after sending SYN and receiving SYN/ACK) with window size 0 (as reported, it is set to 0 by the TCP client when using some proprietary protocol), the ACKs would be incorrectly dropped by the line card due to failure in the TCP check. This issue could be avoided by preventing software from dropping packets that fail in the check, for example, by the following CLI command, re# set interfaces ms-3/0/0 services-options ignore-errors tcp. PR1120079
  • If GRES is performed when Routing Engine scale is more than 5M, All FPC's on new Master Routing Engine will flap. PR1127389
  • On MX Series based line card, multiple modifications of firewall filter might cause lookup chip error and traffic blackhole, following jnh_free error messages could help to identify this issue: messages: fpc1 jnh_free(10212): ERROR [FW/3]:1 Paddr 0x006566a9, addr 0x2566a9, part_type 0call_stack 0x40497574 0x418ffa84 0x41900028 0x418ecf94 0x41861690. PR1131828
  • In Service profile, we have same variable used by filter and CoS, for example in profile RLinternet, variable OutFilter is used by out filter and CoS TCP. This is wrong concept as filter and CoS should have different variables. To fix the issue, we need to do the following: 1. In service profiles, add variable for CoS TCP, for example, in profile RLinternet, add OutTcp for TCP 2. In JSRC, add value for TCP varible OutTcp, for example, ?OutTcp:any="1M"' 3. In JSRC, change value of out filter to distinguish from TCP, for example, 'OutFilter:any="1M-out" 4. Modify out filter name, for example, from "1M" to "1M-out". PR1154982
  • On MX Series routers with MS-MICs and MS-MPCs, the available addresses field in the output of the show services nat pool detail command is always displayed as zero when destination NAT (dNAT) is configured. However, this field displays the correct number of addresses available for allocation when basic NAT or Nnetwork Address Port Translation (NAPT) is configured. PR1158435
  • On MX Series routers with MS-MICs and MS-MPCs with the syslog statement included at the [edit services cos rule rule-name term term-name then] hierarchy level, a system log message is not generated when a CoS rule term is matched, in contrast to the expected behavior in which system log messages are generated when a NAT rule term is matched. PR1159231
  • On MX Series platforms when MPC experiences a FATAL error, it gets reported to the chassisd daemon. Based on the action that is defined for a FATAL error, the chassisd will take subsequent action for the FATAL error. By default, the action for the FATAL error is to reset the MPC. When the MPC reports a FATAL error, chassisd will send an offline message and will power off the MPC upon the ack reception. However if the MPC is in busy state for any reason, the ack does not come in time and hence there would be a delay in bringing down the MPC. The fix of this PR ensures bringing down the MPC in time upon a FATAL error. PR1159742
  • This is an intermittent issue. Assuming that AE is configured with the bypass-queuing-chip configuration statement. Now followup configuration changes are such that removing child link(s) from AE bundle, configuring per-unit-scheduler on the removed child link(s) in a single commit causes intermittent issues with per-unit-scheduler configuration updates to cosd and the Packet Forwarding Engine. Hence, dedicated scheduler nodes might not be created for all units or IFLs. PR1162006
  • Traffic may drop during Routing Engine switchover. PR1164107


  • When performing a Routing Engine switchover (including during ISSU), a ksyncd core might be generated. PR1078894

Interfaces and Chassis

  • After removing a child link from AE bundle, in the output of "show interface <AE> detail", the packets count on the remaining child link spikes, then if we add the previous child link back, the count recovers to normal. PR1091425
  • The outbound PPPoE control packets sourced from PPPoE daemon, such as PADO, error PADS and PADT, are assigned to queue 0 instead of queue 3. PR1154070
  • With MX Series platforms acting as TWAMP client and vMX platform acting as TWAMP server setup, we see probe packet loss at TWAMP server i.e. on VMX with Junos OS version 15.1F5 Release. When TWAMP target interface address is configured as Media interface (-ge/-xe), probe packets are getting dropped at vMx because of ENDIAN conversion of UDP checksum (vMx is Little Endian and Mx is BIG ENDIAN platform) in the probe packet. This issue was seen earlier in Junos OS 15.1F4 release but was resolved through PR#1125516. However, due to some merge issue the fix got overwritten and this issue re-surfaced. Also, when TWAMP target interface address is configured as si- interface we again see probe-packet loss but this time not because of UDP checksum error. Here, issue appears because of some looping issue and packet after getting processed at LU (Timestamped at LU) is not able to go out of media interface. Sometimes enabling some debug logs at Packet Forwarding Engine and changing TWAMP probe packet size resolves issue (but not always). PR1164093
  • When two line-cards are taken offline back to back (without delay between issuing the line-card offline command) we are hitting the issue described in the PR (some planes goes to check state). This issue could be prevented by giving a 1 second delay between the offline commands. Workaround would be to offline and online the planes which are in Check state. PR1164648
  • When running traffic during vMX upgrade and your configuration contains filter-list(s) attached to aggregated ethernet interface(s) and those filter-list(s) have logical-interface-policer(s) riot may reboot constantly. PR1165613

Layer 2 Features

  • On GRES switch of mastership of Routing Engine via request chassis routing-engine master switch, the dot1xd daemon will crash multiple times when 128K IFLs are configured in the MX960 chassis. PR1118475
  • In 15.1R3 with tomcat mode enabled, DHCP subscriber management with IRB interfaces is not reliable. It is possible that the DHCP bindings are unable to fully establish with IRB interfaces due to this reason. However, these bindings with same IRB interfaces should come up properly with tomcat disabled. PR1155502
  • When an MX Series router is acting as DHCP relay to selectively process client traffic with any forward-only configuration, if a downstream device acts as a Layer 2 DHCP relay where it adds an OPTION-82 record but not a giaddr (Gateway Address field), and in addition, the downstream Layer 2 DHCP relay adds the option 82 record in a non-compliant (illegal) way by inserting the OPTION-82 record in front of other existing option records, bad packet format of the DHCP discover/request will send to the server. PR1157800
  • During SIB yanking test, it was found for some corner cases that PF idle insertion setting is incorrect switching from 8 SIB to 9 SIB with the lower idle insertion, then caused some traffic drop and traffic throughput reduction.PR1162977


  • These benign error messages can occur when bringing up new interfaces or line cards and can be ignored. The Junos OS code will be cleaned up to remove these messages in later code. PR1136033
  • When graceful Routing Engine switchover (GRES) is done between the master and backup Routing Engines of different memory capabilities (such that one has only enough memory to run routing protocol process (rpd) in 32-bit mode while the other is capable of 64-bit mode, which could be caused by using Junos OS Release 13.3 onwards with the configuration statement auto-64-bit configured, or, using Junos OS Release 15.1 onwards even without the configuration statement), rpd might crash on the new master Routing Engine. As a workaround, this issue could be avoided by the CLI command set system processes routing force-32-bit. PR1141728
  • MPLS TED might not select random links to calculate the ERO when OSPF is overloaded. Instead, only one or two interfaces will be used for all the configured LSPs originating from the router. PR1147832
  • FPC cores may be seen during AE flap on Type 5-3D FPCs. PR1164175

Platform and Infrastructure

  • The show multicast route extensive command takes more than 2 minutes to display the output when there are more than 1000 routes. PR1084983
  • Junos OS defines SNMP ifXTable (ifJnxInErrors/ifJnxInL3Incompletes) counter as 64-bit width, but it worked as 32-bit width counter. It works as 64-bit width counter after the fix. PR1105266
  • We support maximum of 1024 Softwire concentrators with Inline-6rd. When we configure more than 1024 SCs i.e, 1025 SCs, we see commit error message which is not very informative. PR1153092

Routing Protocols

  • Applying add-path to BGP groups with both IPv4 and IPv6 neighbors causes multiple update groups, as opposed to just one update group per BGP group configuration as is seen when add-path is not configured. PR1144136

Subscriber Management and Services

  • While bringing down subscribers, the system generates [ Deinstantiate Service Failed permanently, daemon: cosd ] error message. PR1136083
  • On MX Series routers with "subscriber-management" enabled, the BBE subscriber management daemon (bbe-smgd) might crash on the backup Routing Engine when performing graceful Routing Engine switchover (GRES) during subscribers concurrent login/logout. PR1147498
  • Subscriber traffic in an LNS coming from the core network is not switched properly when the incoming interface is an irb interface. PR1148533

User Interface and Configuration

  • When committing a configuration with a very long as-path, in this case the as-path is almost 12000 characters long, the commitd process might crash. The commitd process restart results in a minimal impact on the system. As a workaround, please configure as-path to be less than 4096 characters long. PR1119529


  • The fix is to have MVPN walk the suppress list of cmcast entries and try to install the forwarding entries, even if we have moved from 'unsuppress' back to 'suppress'. If there is a PIM installed forwarding entry, then MVPN will be successful in installing the forwarding route. Otherwise, the entry will stay on the suppress list. PR1144207

Related Documentation

Modified: 2016-09-29