Leonardo is removed from Unity's downward facing docking port. Photo: Spaceflight Now/NASA TV

In the end, troubleshooters concluded the shuttle's four general purpose computers, or GPCs, were healthy and that there was no need to order any such electronic brain surgery on a vehicle operating in the hostile environment of space.

The astronauts then pressed ahead with work to close up the Italian-built Leonardo cargo module so it could be detached from the international space station's Unity module and remounted in Discovery's payload bay for return to Earth.

Already running behind schedule because of the computer troubleshooting, the astronauts ran into fresh problems early today with leaky fittings in hoses used to verify pressure integrity after the hatches are closed and the vestibule between the modules is dropped to vacuum.

The fittings in question were tightened and the crew did a second vestibule depressurization. While a bit of leakage remained, engineers were confident it was limited to the pressure testing system and that the seals were, in fact, tight.

The shuttle's robot arm takes a firm grip on the Leonardo module. Photo: Spaceflight Now/NASA TV

"Although the hosing itself showed some leakage, we were comfortable that we had a good seal so we went ahead and gave the shuttle crew a go to start unberthing (Leonardo)," said station flight director Rick La Brode.

When all was said and done, the crew was running some five hours behind schedule when astronaut Andrew Thomas, operating Discovery's robot arm, finally detached Leonardo from the space station.

He then mounted the module in the shuttle's cargo bay for return to Earth without incident, setting the stage for Discovery's undocking at 11:32 p.m. EST tonight.

Loaded with nearly five tons of equipment and supplies, Leonardo was launched to the station aboard Discovery and attached the Unity module's downward facing port last Monday.

Space station commander Bill Shepherd and Expedition 2 crew member Jim Voss close the hatch to the Leonardo module. Photo: Spaceflight Now/NASA TV

Work to unload the cargo carrier went smoothly, but it took longer than expected to repack it with about one ton of station trash, dirty clothes, computer disks, discarded equipment and other items. As a result, Discovery's mission was extended one day to give the crew more time.

Despite the delays getting Leonardo back into Discovery's cargo bay today, lead flight director John Shannon said the crew was able to complete preliminary undocking preparations overnight as planned.

"Right now, we anticipate they'll get to bed on time and we'll do the undocking as planned," Shannon said.

The computer issue caught flight controllers - and the crew - by surprise. Given the critical nature of the system - the four GPCs control all aspects of shuttle operation - mission managers left no stone unturned to verify their health.

The shuttle is equipped with four general purpose computers that run identical software. The computers "vote" on critical operations to provide multiple levels of redundancy. A fifth backup computer also is on board, this one loaded with software written by a different vendor in case of a common bug that might sideline all four GPCs.

Discovery's problem developed Saturday morning, before the crew went to bed, when flight controllers noticed lower-than-desireable temperatures in a Freon coolant loop. At the time, the shuttle's avionics system was in a so-called "group B powerdown" to conserve electricity.

To generate a bit of additional heat for the overly cold Freon loop, the astronauts were asked to power up GPCs 2 and 4, which were in a sort of standby mode as part of the group B powerdown.

On power up, the computers download software and timing information from the GPCs that are alrady operating to ensure all the machines are properly synchronized.

The normal procedure is to activate dormant computers at least 10 seconds apart to make sure they don't attempt to access and download data from the operating computers at the same time. In that case, data can become corrupted.

But it appears the crew reactivated GPCs 2 and 4 within about six seconds of each other and engineers spent the day evaluating the flight computers and their software to determine what, if anything, might need to be done.

Shannon blamed the glitch on flight controllers, saying the crew was not advised to wait 10 seconds between powering up GPCs 2 and 4.

"The ground called to turn those two GPCs on and they did not reference any procedure," he said. "If they had, the 10-second warning would have been in there. The crew did exactly what the ground told them and when we went back and looked we said, ah, we should have given them the warning to wait 10 seconds."

Engineers were initially concerned they might have to order the crew to reload the software in all four general purpose computers from scratch using the backup flight system, or BFS, computer.

While there have been computer problems in the past, no crew in the 20-year, 103-flight history of the shuttle program has ever had to reload the main flight control software during a mission.

The International Space Station towers over shuttle Discovery. Photo: Spaceflight Now/NASA TV

As it turned out, an engineering review determined that the GPCs can be started up within five seconds of each other without causing any problems. The 10-second rule was intended to provide a bit of margin.

After meeting for more than an hour Saturday night to assess engineering data, NASA's mission management team ordered a special test of the computers to provide additional insight into their condition.

The crew was asked to switch the GPCs to a different operating mode - OPS-8 flight control system checkout software - and then back again to OPS-2 software, the computer programs that control normal operations in orbit.

If the computers were healthy, the thinking went, the machines would shift between modes without any problems. And to the relief of the data processing system officer in mission control, that's exactly what they did.

"Houston, Discovery, we show ourselves complete," commander James Wetherbee radioed around 11:30 p.m. "Thank you very much for all the analysis and the work overnight and to the MMT also and thanks to Terry for watching over our shoulders. Appreciate it."

"And Discovery, Houston, congratulations on a super job of recovering the GPCs," astronaut Gerhard Thiele replied from mission control. "We have full confidence that things will work from now on. For the big picture for the rest of today, we're looking to the station side of the hatch to get ready for MPLM demate and putting it back into the cargo bay."