Why did my HP BladeSystem server with HP 650FLB or 650M adapters connected to HP Virtual Connect switches loose access to all LUNs from the SAN?

Issue

• One of the blade servers momentarily lost link to the SAN on both HBA ports:

Mar 26 20:02:00 hostname kernel: lpfc 0000:04:00.3: 1:1305 Link Down Event x2 received Data: x2 x20 x800110 x0 x0
Mar 26 20:02:00 hostname kernel: lpfc 0000:04:00.2: 0:1305 Link Down Event x2 received Data: x2 x20 x800110 x0 x0

• Link was regained immediately:

Mar 26 20:02:00 hostname kernel: lpfc 0000:04:00.3: 1:1303 Link Up Event x4 received Data: x4 x0 x40 x0 x0 x0 0
Mar 26 20:02:00 hostname kernel: lpfc 0000:04:00.3: 1:2718 Clear Virtual Link Received for VPI 0x100 tag 0x5
Mar 26 20:02:00 hostname kernel: lpfc 0000:04:00.2: 0:1303 Link Up Event x4 received Data: x4 x0 x40 x0 x0 x0 0
Mar 26 20:02:00 hostname kernel: lpfc 0000:04:00.2: 0:2718 Clear Virtual Link Received for VPI 0x0 tag 0x5

• But the server still lost access to all the SAN paths:

Mar 26 20:02:05 hostname multipathd: 2001738002abe0239: remaining active paths: 0
Mar 26 20:02:06 hostname multipathd: 2001738002abe0227: remaining active paths: 0
Mar 26 20:02:06 hostname multipathd: 2001738002abe022a: remaining active paths: 0
Mar 26 20:02:06 hostname multipathd: 2001738002abe023c: remaining active paths: 0
[...]

• Why is access to the LUNs not recovered?