Post upgrade to JUNOS 21.4R3-S4.9 we could see the following log messages login continuously.-

[Mon Nov 20 11:56:08.178 LOG: Err] WRITE2 failed addr 50 data 0
[Mon Nov 20 11:56:08.178 LOG: Err] [RERROR]pfpga_pcie_i2c_read_delay:169: i2c read error addr: 0x50 seg: 0x41 off: 0x0 len: 0x20
[Mon Nov 20 11:56:16.328 LOG: Err] WRITE2 failed addr 50 data 0
[Mon Nov 20 11:56:16.328 LOG: Err] [RERROR]pfpga_pcie_i2c_read_delay:169: i2c read error addr: 0x50 seg: 0x41 off: 0x0 len: 0x20
[Mon Nov 20 11:57:16.777 LOG: Err] WRITE2 failed addr 50 data 0
[Mon Nov 20 11:57:16.777 LOG: Err] [RERROR]pfpga_pcie_i2c_read_delay:169: i2c read error addr: 0x50 seg: 0x41 off: 0x0 len: 0x20
[Mon Nov 20 11:57:49.609 LOG: Info]
INTR: throttle 60sec PECHIP[0]:pe.irw.intr.status:do_not_transmit_in(0): (Count:1)
[Mon Nov 20 11:59:47.943 LOG: Err] WRITE2 failed addr 50 data 11
[Mon Nov 20 11:59:47.943 LOG: Err] [RERROR]pfpga_pcie_i2c_read_delay:169: i2c read error addr: 0x50 seg: 0x47 off: 0x0 len: 0x20
[Mon Nov 20 12:00:22.034 LOG: Err] WRITE2 failed addr 50 data 0
[Mon Nov 20 12:00:22.034 LOG: Err] [RERROR]pfpga_pcie_i2c_read_delay:169: i2c read error addr: 0x50 seg: 0x45 off: 0x0 len: 0x20

• Issue only seen intermittently on Lumentum CWDM4 100G XCVRs.
• The i2c read failure is seen randomly and does not seem to have any functional impact.
• The Lumentum vendor XCVR takes more time to respond between successive reads and sometimes reads fail. So, increasing the time between successive reads is the key here.
• Keeping the delay too high will create delay in optics operation. Keeping the delay too low will bring back the i2c read error messages.
• After trying a lot of values, Engineering team has arrived at an optimal value of 170 usec with which errors are not seen.
• We have found the fix for this issue. But unfortunately, we will not be able to get it into our build 22.4R3.
• As mentioned earlier, these logs are not traffic-impacting. You may ask customer to whitelist these log signatures.