Transparent server failover in stateful distributed protocols

A non-transitory computer-readable medium storing instructions, which when executed by a processing device, cause the processing device to perform a method comprising:

• intercepting, by a proxy, a plurality of requests sent by a client and addressed to a first server;
• adding, by the proxy, an entry for each of the plurality of requests to a data structure, each entry including data from one of the plurality of requests;
• forwarding, by the proxy, each of the plurality of requests to the first server;
• detecting, by the proxy, that the first server is unavailable;
• connecting, by the proxy in response to detecting the first server was unavailable, to the first server when the first server recovers or to a second server serving as a backup for the first server;
• transmitting, by the proxy to the first or second server, one or more commands using the entries of the data structure to restore state information for the client;
• and
• transmitting, by the proxy to the first or second server, an indication that state recovery for the client is complete.

An apparatus comprising:

• a processing device;
• and
• a memory coupled to the processing device, the memory storing instructions which, when executed by the processing device, cause the apparatus to:
• intercept a plurality of requests sent by a client and addressed to a first server;
• add an entry for each of the plurality of requests to a data structure, each entry including data from one of the plurality of requests;
• forward each of the plurality of requests to the first server;
• detect that the first server is unavailable;
• in response to detecting the first server was unavailable, connect to the first server when the first server recovers or to a second server serving as a backup for the first server;
• transmit, to the first or second server, one or more commands using the entries of the data structure to restore state information for the client;
• and
• transmit, by the proxy to the first or second server, an indication that state recovery for the client is complete.