== Description of problem:
When the interface is not managed by NetworkManager the IPsrcaddr resource fails
to stop after successful IPsrcadd start and monitor.

== Version-Release number of selected component (if applicable):
resource-agents-3.9.5-105.el7 

== How reproducible:
Always

== Steps to Reproduce:
1. Configure interface ensX with IPADDR, GATEWAY and NM_CONTROLLED="no" (for example IPADDR=10.0.0.85, PREFIX=24, GATEWAY=10.0.0.1)
2. configure IPaddr2 resource (vip) with IP y.y.y.y in cluster with same subnet as the ensX that is different from IPADDR and GATEWAY (for example y.y.y.y=10.0.0.10)
3. Configure IPsrcaddr resource (vip_route) in cluster to use the IPaddr2 as default route (ipaddress=10.0.0.10 cidr_netmask=24)
4. Disable both resources and try debug-start,debug-monitor, debug-stop them on single node
  # pcs resource disable vip
  # pcs resource disable vip_route
  # pcs resource debug-start vip
  # pcs resource debug-start vip_route
  # pcs resource debug-monitor vip_route
  # pcs resource debug-stop vip_route

== Actual results:
  # pcs resource debug-start vip
  Operation start for vip (ocf:heartbeat:IPaddr2) returned 0

  # pcs resource debug-start vip_route
  Operation start for vip_route (ocf:heartbeat:IPsrcaddr) returned 0

  # pcs resource debug-monitor vip_route
  Operation monitor for vip_route (ocf:heartbeat:IPsrcaddr) returned 0

  # pcs resource debug-stop vip_route
Error performing operation: Operation not permitted
Operation stop for vip_route (ocf:heartbeat:IPsrcaddr) returned 1
 >  stderr: Usage: ip route { list | flush } SELECTOR
 >  stderr:        ip route save SELECTOR
 >  stderr:        ip route restore
 >  stderr:        ip route showdump
 >  stderr:        ip route get ADDRESS [ from ADDRESS iif STRING ]
 >  stderr:                             [ oif STRING ]  [ tos TOS ]
 >  stderr:                             [ mark NUMBER ]
 >  stderr:        ip route { add | del | change | append | replace } ROUTE
 >  stderr: SELECTOR := [ root PREFIX ] [ match PREFIX ] [ exact PREFIX ]
 >  stderr:             [ table TABLE_ID ] [ proto RTPROTO ]
 >  stderr:             [ type TYPE ] [ scope SCOPE ]
 >  stderr: ROUTE := NODE_SPEC [ INFO_SPEC ]
 >  stderr: NODE_SPEC := [ TYPE ] PREFIX [ tos TOS ]
 >  stderr:              [ table TABLE_ID ] [ proto RTPROTO ]
 >  stderr:              [ scope SCOPE ] [ metric METRIC ]
 >  stderr: INFO_SPEC := NH OPTIONS FLAGS [ nexthop NH ]...
 >  stderr: NH := [ via ADDRESS ] [ dev STRING ] [ weight NUMBER ] NHFLAGS
 >  stderr: OPTIONS := FLAGS [ mtu NUMBER ] [ advmss NUMBER ]
 >  stderr:            [ rtt TIME ] [ rttvar TIME ] [reordering NUMBER ]
 >  stderr:            [ window NUMBER ] [ cwnd NUMBER ] [ initcwnd NUMBER ]
 >  stderr:            [ ssthresh NUMBER ] [ realms REALM ] [ src ADDRESS ]
 >  stderr:            [ rto_min TIME ] [ hoplimit NUMBER ] [ initrwnd NUMBER ]
 >  stderr:            [ features FEATURES ] [ quickack BOOL ] [ congctl NAME ]
 >  stderr:            [ expires TIME ]
 >  stderr: TYPE := { unicast | local | broadcast | multicast | throw |
 >  stderr:           unreachable | prohibit | blackhole | nat }
 >  stderr: TABLE_ID := [ local | main | default | all | NUMBER ]
 >  stderr: SCOPE := [ host | link | global | NUMBER ]
 >  stderr: NHFLAGS := [ onlink | pervasive ]
 >  stderr: RTPROTO := [ kernel | boot | static | NUMBER ]
 >  stderr: TIME := NUMBER[s|ms]
 >  stderr: BOOL := [1|0]
 >  stderr: FEATURES := ecn
 >  stderr: ocf-exit-reason:command 'ip route replace  dev ensX' failed

== Expected results:
  # pcs resource debug-start vip
  Operation start for vip (ocf:heartbeat:IPaddr2) returned 0

  # pcs resource debug-start vip_route
  Operation start for vip_route (ocf:heartbeat:IPsrcaddr) returned 0

  # pcs resource debug-monitor vip_route
  Operation monitor for vip_route (ocf:heartbeat:IPsrcaddr) returned 0

  # pcs resource debug-stop vip_route
  Operation stop for vip_route (ocf:heartbeat:IPsrcaddr) returned 0

== Additional info:
Issue is not reproducible when the interface is managed by NetworkManager.
Below are outputs from 'ip route' command when NetworkManager is and is not in use.

=======
## bond0 without NetworkManager clean start
  default via 10.0.0.1 dev bond0
  10.0.0.0/24 dev bond0 proto kernel scope link src 10.0.0.85

  # after starting IPsrcaddr
  default via 10.0.0.1 dev bond0 src 10.0.0.10
  10.0.0.0/24 dev bond0 scope link src 10.0.0.10

  # attempting to stop IPsrcaddr results in error (exit code 1)

## bond0 with NetworkManager clean start
  default via 10.0.0.1 dev bond0 proto static metric 300  
  10.0.0.0/24 dev bond0 proto kernel scope link src 10.0.0.85 metric 300

  # after starting IPsrcaddr
  default via 10.0.0.1 dev bond0 proto static src 10.0.0.10 metric 300
  10.0.0.0/24 dev bond0 scope link src 10.0.0.10
  10.0.0.0/24 dev bond0 proto kernel scope link src 10.0.0.85 metric 300

  # after stopping the IPsrcaddr
  default via 10.0.0.1 dev bond0 proto static metric 300
  10.0.0.0/24 dev bond0 scope link
  10.0.0.0/24 dev bond0 proto kernel scope link src 10.0.0.85 metric 300
=======

Customer provided us with the patch containing workaround that works in their environment (there are some hardcoded things)

--- /usr/lib/ocf/resource.d/heartbeat/IPsrcaddr 2017-06-23 09:32:28.000000000 +0200
+++ /usr/lib/ocf/resource.d/heartbeat/IPsrcaddr.custom 2017-10-16 15:55:31.617636470 +0200
@@ -172,7 +172,7 @@
                rc=$OCF_SUCCESS
                ocf_log info "The ip route has been already set.($NETWORK, $INTERFACE, $ROUTE_WO_SRC)"
        else
-               ip route replace $NETWORK dev $INTERFACE src $1 || \
+               ip route replace $NETWORK dev $INTERFACE proto kernel scope link src $1 || \
                        errorexit "command 'ip route replace $NETWORK dev $INTERFACE src $1' failed"
 
                $CMDCHANGE $ROUTE_WO_SRC src $1 || \
@@ -204,7 +204,10 @@
          
        [ $rc = 2 ] && errorexit "The address you specified to stop does not match the preferred source address"
 
-       ip route replace $NETWORK dev $INTERFACE || \
+# WORKAROUND !!!!
+       PRIMARY=`ip -4 addr show bond1 primary | fgrep inet | awk -c '{ print $2; }' | cut -f1 -d\/`
+
+       ip route replace $NETWORK dev $INTERFACE proto kernel scope link src ${PRIMARY} || \
                errorexit "command 'ip route replace $NETWORK dev $INTERFACE' failed"
 
        $CMDCHANGE $ROUTE_WO_SRC || \


Additional tests:
- test when NetworkManager service is completely off
  - no change in behaviour (systemctl disable NetworkManager and reboot, verified that NM was stopped)
- test adding the same route as NetworkManager adds when it is used to see if then this works
  - adding a route _after_ starting IPsrcaddr gets us into state from which things starts to work correctly.
    However we cannot add the same route _before_ starting the IPsrcaddr as the start of IPsrcaddr would then fail

The following commands seems to causing the failure
207         ip route replace $NETWORK dev $INTERFACE || \

471 INTERFACE=`echo $findif_out | awk '{print $1}'`
472 NETWORK=`ip route list dev $INTERFACE scope link proto kernel match $ipaddress|grep -o '^[^ ]*'`

- In scenario when it fails we see following
>  stderr: + 17:02:43: srca_stop:207: ip route replace dev ens7

- In scenario when it works we see following
>  stderr: + 17:03:05: srca_stop:207: ip route replace 10.0.0.0/24 dev ens7

So it seems that we are missing the $NETWORK. Below are the outputs how the NETWORK is determined
# ip route
default via 10.0.0.1 dev ens7 src 10.0.0.10 
10.0.0.0/24 dev ens7 scope link src 10.0.0.10 
...
 >  stderr: ++ 17:02:43: 472: ip route list dev ens7 scope link proto kernel match 10.0.0.10
 >  stderr: ++ 17:02:43: 472: grep -o '^[^ ]*'
 >  stderr: + 17:02:43: 472: NETWORK=

# ip route
default via 10.0.0.1 dev ens7 src 10.0.0.10 
10.0.0.0/24 dev ens7 scope link src 10.0.0.10 
10.0.0.0/24 dev ens7 proto kernel scope link src 10.0.0.85 metric 100 
...
 >  stderr: ++ 17:08:13: 472: ip route list dev ens7 scope link proto kernel match 10.0.0.10
 >  stderr: ++ 17:08:13: 472: grep -o '^[^ ]*'
 >  stderr: + 17:08:13: 472: NETWORK=10.0.0.0/24

So it looks that in some cases we fail to detect correctly the NETWORK